├── README.md
├── code
    ├── .ipynb_checkpoints
    │   ├── clean_data-checkpoint.ipynb
    │   ├── combine_fetures-checkpoint.ipynb
    │   ├── extract_fetures-checkpoint.ipynb
    │   ├── holiday_crawl-checkpoint.ipynb
    │   ├── predict_result-checkpoint.ipynb
    │   ├── rules-checkpoint.ipynb
    │   ├── run-checkpoint.ipynb
    │   ├── split_samples-checkpoint.ipynb
    │   ├── train_model-checkpoint.ipynb
    │   ├── trick_extract-checkpoint.ipynb
    │   └── visualize-checkpoint.ipynb
    ├── clean_data.ipynb
    ├── clean_data.py
    ├── combine_fetures.ipynb
    ├── extract_fetures.ipynb
    ├── extract_fetures.py
    ├── holiday_crawl.ipynb
    ├── predict_result.ipynb
    ├── result.txt
    ├── rules.ipynb
    ├── run.ipynb
    ├── split_samples.ipynb
    ├── train_model.ipynb
    ├── trick_extract.ipynb
    ├── trick_extract.py
    ├── visualize.ipynb
    └── weather_crawl.py
└── dataset
    ├── holiday.csv
    ├── train.csv
    └── yangzhong.csv


/README.md:
--------------------------------------------------------------------------------
 1 | # 管道搭建示例
 2 | ## 基于天池 大航杯"智造扬中”电力AI (失败解决方案)
 3 | 
 4 | 详细请见[博文](http://blog.zhanglun.me/2017/06/13/%E5%A4%A7%E8%88%AA%E6%9D%AF%E2%80%9C%E6%99%BA%E9%80%A0%E6%89%AC%E4%B8%AD%E2%80%9D%E7%94%B5%E5%8A%9BAI%E5%A4%A7%E8%B5%9B%E5%8F%82%E8%B5%9B%E7%BB%8F%E9%AA%8C/) 
 5 | 
 6 | #### 文件夹介绍
 7 | 
 8 | code  --主要代码文件
 9 | 
10 | dataset  --数据文件
11 | 
12 | 
13 | 项目依赖:
14 | python3
15 | numpy 
16 | pandas 
17 | scipy
18 | sklearn
19 | mlxtend


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/clean_data-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import sys\n",
 12 |     "from datetime import datetime, timedelta, date\n",
 13 |     "\n",
 14 |     "import numpy as np\n",
 15 |     "import pandas as pd\n",
 16 |     "import scipy as sp\n",
 17 |     "\n",
 18 |     "from sklearn.ensemble import IsolationForest"
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": 2,
 24 |    "metadata": {
 25 |     "collapsed": false
 26 |    },
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "if sys.argv[1] == 'test':\n",
 30 |     "    is_train = False\n",
 31 |     "    train_path = '../dataset/fetures/test.csv'\n",
 32 |     "else:\n",
 33 |     "    is_train = True\n",
 34 |     "    if sys.argv[2].endswith('.json'):\n",
 35 |     "        month = '9'\n",
 36 |     "    else:\n",
 37 |     "        month = sys.argv[2]\n",
 38 |     "    train_path = '../dataset/fetures/{}/train.csv'.format(month)"
 39 |    ]
 40 |   },
 41 |   {
 42 |    "cell_type": "code",
 43 |    "execution_count": 3,
 44 |    "metadata": {
 45 |     "collapsed": false
 46 |    },
 47 |    "outputs": [],
 48 |    "source": [
 49 |     "train = pd.read_csv(train_path, parse_dates=['record_date'])"
 50 |    ]
 51 |   },
 52 |   {
 53 |    "cell_type": "code",
 54 |    "execution_count": 4,
 55 |    "metadata": {
 56 |     "collapsed": false
 57 |    },
 58 |    "outputs": [
 59 |     {
 60 |      "name": "stdout",
 61 |      "output_type": "stream",
 62 |      "text": [
 63 |       "<class 'pandas.core.frame.DataFrame'>\n",
 64 |       "RangeIndex: 532156 entries, 0 to 532155\n",
 65 |       "Data columns (total 3 columns):\n",
 66 |       "record_date          532156 non-null datetime64[ns]\n",
 67 |       "user_id              532156 non-null int64\n",
 68 |       "power_consumption    532156 non-null int64\n",
 69 |       "dtypes: datetime64[ns](1), int64(2)\n",
 70 |       "memory usage: 12.2 MB\n"
 71 |      ]
 72 |     }
 73 |    ],
 74 |    "source": [
 75 |     "train.info()"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 5,
 81 |    "metadata": {
 82 |     "collapsed": true
 83 |    },
 84 |    "outputs": [],
 85 |    "source": [
 86 |     "threshold = 0.95"
 87 |    ]
 88 |   },
 89 |   {
 90 |    "cell_type": "code",
 91 |    "execution_count": 6,
 92 |    "metadata": {
 93 |     "collapsed": true
 94 |    },
 95 |    "outputs": [],
 96 |    "source": [
 97 |     "def clean_quantile(df):\n",
 98 |     "    up = df.power_consumption.quantile(threshold)\n",
 99 |     "    down = df.power_consumption.quantile(1-threshold)\n",
100 |     "    mean = df.power_consumption.mean()\n",
101 |     "    print(df['user_id'].iloc[0], 'up:', up, 'down:', down, 'mean:', mean, end='\\n', file=open('../dataset/clean.txt', 'a+'))\n",
102 |     "    df.loc[(df.power_consumption>=up)|(df.power_consumption<=down), 'power_consumption'] = df.power_consumption.mean()\n",
103 |     "    return df"
104 |    ]
105 |   },
106 |   {
107 |    "cell_type": "code",
108 |    "execution_count": 7,
109 |    "metadata": {
110 |     "collapsed": false
111 |    },
112 |    "outputs": [],
113 |    "source": [
114 |     "train = train.groupby('user_id', as_index=True).apply(clean_quantile)"
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "markdown",
119 |    "metadata": {},
120 |    "source": [
121 |     "# 去除这些公司后结果变差,可能是因为未来也会造成影响"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 9,
127 |    "metadata": {
128 |     "collapsed": true
129 |    },
130 |    "outputs": [],
131 |    "source": [
132 |     "def clean_all_zero(df):\n",
133 |     "    ndf = df.loc[df.power_consumption!=1]\n",
134 |     "    if not ndf.empty:\n",
135 |     "        return df\n",
136 |     "    else:\n",
137 |     "        print(df['user_id'].iloc[0])\n",
138 |     "        return ndf\n"
139 |    ]
140 |   },
141 |   {
142 |    "cell_type": "code",
143 |    "execution_count": 10,
144 |    "metadata": {
145 |     "collapsed": true
146 |    },
147 |    "outputs": [
148 |     {
149 |      "name": "stdout",
150 |      "output_type": "stream",
151 |      "text": [
152 |       "259\n",
153 |       "413\n",
154 |       "418\n",
155 |       "462\n",
156 |       "469\n",
157 |       "522\n",
158 |       "533\n",
159 |       "549\n",
160 |       "550\n",
161 |       "551\n",
162 |       "553\n",
163 |       "557\n",
164 |       "570\n",
165 |       "586\n",
166 |       "588\n",
167 |       "589\n",
168 |       "598\n",
169 |       "606\n",
170 |       "609\n",
171 |       "617\n",
172 |       "622\n",
173 |       "633\n",
174 |       "634\n",
175 |       "637\n",
176 |       "638\n",
177 |       "639\n",
178 |       "645\n",
179 |       "646\n",
180 |       "647\n",
181 |       "650\n",
182 |       "651\n",
183 |       "652\n",
184 |       "653\n",
185 |       "654\n",
186 |       "655\n",
187 |       "656\n",
188 |       "657\n",
189 |       "658\n",
190 |       "661\n",
191 |       "663\n",
192 |       "664\n",
193 |       "665\n",
194 |       "666\n",
195 |       "667\n",
196 |       "668\n",
197 |       "669\n",
198 |       "670\n",
199 |       "671\n",
200 |       "672\n",
201 |       "673\n",
202 |       "674\n",
203 |       "675\n",
204 |       "676\n",
205 |       "677\n",
206 |       "678\n",
207 |       "679\n",
208 |       "680\n",
209 |       "681\n",
210 |       "682\n",
211 |       "683\n",
212 |       "684\n",
213 |       "685\n",
214 |       "686\n",
215 |       "687\n",
216 |       "688\n",
217 |       "689\n",
218 |       "690\n",
219 |       "691\n",
220 |       "692\n",
221 |       "693\n",
222 |       "694\n",
223 |       "695\n",
224 |       "696\n",
225 |       "709\n",
226 |       "891\n",
227 |       "950\n",
228 |       "1044\n",
229 |       "1045\n",
230 |       "1111\n",
231 |       "1129\n",
232 |       "1233\n",
233 |       "1234\n",
234 |       "1261\n",
235 |       "1280\n",
236 |       "1413\n",
237 |       "<class 'pandas.core.frame.DataFrame'>\n",
238 |       "RangeIndex: 501046 entries, 0 to 501045\n",
239 |       "Data columns (total 3 columns):\n",
240 |       "record_date          501046 non-null datetime64[ns]\n",
241 |       "user_id              501046 non-null int64\n",
242 |       "power_consumption    501046 non-null float64\n",
243 |       "dtypes: datetime64[ns](1), float64(1), int64(1)\n",
244 |       "memory usage: 11.5 MB\n"
245 |      ]
246 |     }
247 |    ],
248 |    "source": [
249 |     "# train = train.groupby('user_id').apply(clean_all_zero).reset_index(drop=True)\n",
250 |     "# train.info()"
251 |    ]
252 |   },
253 |   {
254 |    "cell_type": "code",
255 |    "execution_count": null,
256 |    "metadata": {
257 |     "collapsed": true
258 |    },
259 |    "outputs": [],
260 |    "source": [
261 |     "# test = test.groupby('user_id').apply(clean_all_zero).reset_index(drop=True)\n",
262 |     "# test.info()"
263 |    ]
264 |   },
265 |   {
266 |    "cell_type": "markdown",
267 |    "metadata": {},
268 |    "source": [
269 |     "# 添加公司用电量平滑处理"
270 |    ]
271 |   },
272 |   {
273 |    "cell_type": "code",
274 |    "execution_count": 11,
275 |    "metadata": {
276 |     "collapsed": true
277 |    },
278 |    "outputs": [],
279 |    "source": [
280 |     "def rolling_power_consumption(df):\n",
281 |     "    ndf = df.set_index('record_date')\n",
282 |     "    ndf = ndf.rolling(2).mean()\n",
283 |     "    return ndf.reset_index()"
284 |    ]
285 |   },
286 |   {
287 |    "cell_type": "code",
288 |    "execution_count": 12,
289 |    "metadata": {
290 |     "collapsed": false
291 |    },
292 |    "outputs": [],
293 |    "source": [
294 |     "train = train.groupby('user_id').apply(rolling_power_consumption).dropna(subset=['user_id']).reset_index(drop=True)"
295 |    ]
296 |   },
297 |   {
298 |    "cell_type": "code",
299 |    "execution_count": 13,
300 |    "metadata": {
301 |     "collapsed": false
302 |    },
303 |    "outputs": [
304 |     {
305 |      "name": "stdout",
306 |      "output_type": "stream",
307 |      "text": [
308 |       "<class 'pandas.core.frame.DataFrame'>\n",
309 |       "RangeIndex: 499677 entries, 0 to 499676\n",
310 |       "Data columns (total 3 columns):\n",
311 |       "record_date          499677 non-null datetime64[ns]\n",
312 |       "user_id              499677 non-null float64\n",
313 |       "power_consumption    499677 non-null float64\n",
314 |       "dtypes: datetime64[ns](1), float64(2)\n",
315 |       "memory usage: 11.4 MB\n"
316 |      ]
317 |     }
318 |    ],
319 |    "source": [
320 |     "train.to_csv(train_path, index=False)\n",
321 |     "train.info()"
322 |    ]
323 |   }
324 |  ],
325 |  "metadata": {
326 |   "anaconda-cloud": {},
327 |   "hide_input": false,
328 |   "kernelspec": {
329 |    "display_name": "Python 3",
330 |    "language": "python",
331 |    "name": "python3"
332 |   },
333 |   "language_info": {
334 |    "codemirror_mode": {
335 |     "name": "ipython",
336 |     "version": 3
337 |    },
338 |    "file_extension": ".py",
339 |    "mimetype": "text/x-python",
340 |    "name": "python",
341 |    "nbconvert_exporter": "python",
342 |    "pygments_lexer": "ipython3",
343 |    "version": "3.5.1"
344 |   }
345 |  },
346 |  "nbformat": 4,
347 |  "nbformat_minor": 2
348 | }
349 | 


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/holiday_crawl-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 54,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import requests\n",
 12 |     "import pandas as pd\n",
 13 |     "import json\n",
 14 |     "from os.path import exists"
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "code",
 19 |    "execution_count": 59,
 20 |    "metadata": {
 21 |     "collapsed": true
 22 |    },
 23 |    "outputs": [],
 24 |    "source": [
 25 |     "def write_to_file(filename, data):\n",
 26 |     "    if not exists(filename):\n",
 27 |     "        with open(filename, 'w') as f:\n",
 28 |     "            f.write('date,holiday\\n')   \n",
 29 |     "    for k, v in json.loads(data).items():\n",
 30 |     "        if type(v) == str:\n",
 31 |     "            v = int(v)\n",
 32 |     "        with open(filename, 'a') as f:\n",
 33 |     "            f.write('{},{}\\n'.format(k, v))        "
 34 |    ]
 35 |   },
 36 |   {
 37 |    "cell_type": "code",
 38 |    "execution_count": 2,
 39 |    "metadata": {
 40 |     "collapsed": false
 41 |    },
 42 |    "outputs": [],
 43 |    "source": [
 44 |     "data = pd.date_range('2015/1/1', '2016/12/31')"
 45 |    ]
 46 |   },
 47 |   {
 48 |    "cell_type": "code",
 49 |    "execution_count": 3,
 50 |    "metadata": {
 51 |     "collapsed": false
 52 |    },
 53 |    "outputs": [],
 54 |    "source": [
 55 |     "date_ags = data.map(lambda x:x.strftime('%Y%m%d'))"
 56 |    ]
 57 |   },
 58 |   {
 59 |    "cell_type": "code",
 60 |    "execution_count": 19,
 61 |    "metadata": {
 62 |     "collapsed": true
 63 |    },
 64 |    "outputs": [],
 65 |    "source": [
 66 |     "url_fmt = 'http://tool.bitefu.net/jiari/?d={}'"
 67 |    ]
 68 |   },
 69 |   {
 70 |    "cell_type": "code",
 71 |    "execution_count": 30,
 72 |    "metadata": {
 73 |     "collapsed": false
 74 |    },
 75 |    "outputs": [],
 76 |    "source": [
 77 |     "agrv = ','.join(date_ags[:100])"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": 31,
 83 |    "metadata": {
 84 |     "collapsed": false
 85 |    },
 86 |    "outputs": [
 87 |     {
 88 |      "data": {
 89 |       "text/plain": [
 90 |        "'http://tool.bitefu.net/jiari/?d=20150101,20150102,20150103,20150104,20150105,20150106,20150107,20150108,20150109,20150110,20150111,20150112,20150113,20150114,20150115,20150116,20150117,20150118,20150119,20150120,20150121,20150122,20150123,20150124,20150125,20150126,20150127,20150128,20150129,20150130,20150131,20150201,20150202,20150203,20150204,20150205,20150206,20150207,20150208,20150209,20150210,20150211,20150212,20150213,20150214,20150215,20150216,20150217,20150218,20150219,20150220,20150221,20150222,20150223,20150224,20150225,20150226,20150227,20150228,20150301,20150302,20150303,20150304,20150305,20150306,20150307,20150308,20150309,20150310,20150311,20150312,20150313,20150314,20150315,20150316,20150317,20150318,20150319,20150320,20150321,20150322,20150323,20150324,20150325,20150326,20150327,20150328,20150329,20150330,20150331,20150401,20150402,20150403,20150404,20150405,20150406,20150407,20150408,20150409,20150410'"
 91 |       ]
 92 |      },
 93 |      "execution_count": 31,
 94 |      "metadata": {},
 95 |      "output_type": "execute_result"
 96 |     }
 97 |    ],
 98 |    "source": [
 99 |     "url = url_fmt.format(agrv)\n",
100 |     "url"
101 |    ]
102 |   },
103 |   {
104 |    "cell_type": "code",
105 |    "execution_count": 63,
106 |    "metadata": {
107 |     "collapsed": false
108 |    },
109 |    "outputs": [],
110 |    "source": [
111 |     "for i in range(0, date_ags.shape[0], 100):\n",
112 |     "    agrv = ','.join(date_ags[i:(i+100)])\n",
113 |     "    url = url_fmt.format(agrv)\n",
114 |     "    res = requests.get(url)\n",
115 |     "    write_to_file('../dataset/holiday.csv', res.text)"
116 |    ]
117 |   }
118 |  ],
119 |  "metadata": {
120 |   "anaconda-cloud": {},
121 |   "hide_input": false,
122 |   "kernelspec": {
123 |    "display_name": "Python 3",
124 |    "language": "python",
125 |    "name": "python3"
126 |   },
127 |   "language_info": {
128 |    "codemirror_mode": {
129 |     "name": "ipython",
130 |     "version": 3
131 |    },
132 |    "file_extension": ".py",
133 |    "mimetype": "text/x-python",
134 |    "name": "python",
135 |    "nbconvert_exporter": "python",
136 |    "pygments_lexer": "ipython3",
137 |    "version": "3.5.1"
138 |   }
139 |  },
140 |  "nbformat": 4,
141 |  "nbformat_minor": 2
142 | }
143 | 


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/predict_result-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [
 10 |     {
 11 |      "name": "stderr",
 12 |      "output_type": "stream",
 13 |      "text": [
 14 |       "/home/zhanglun/.pyenv/versions/anaconda3-4.1.0/lib/python3.5/site-packages/sklearn/cross_validation.py:44: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
 15 |       "  \"This module will be removed in 0.20.\", DeprecationWarning)\n"
 16 |      ]
 17 |     }
 18 |    ],
 19 |    "source": [
 20 |     "import pickle\n",
 21 |     "\n",
 22 |     "import numpy as np\n",
 23 |     "import scipy as sp\n",
 24 |     "import pandas as pd\n",
 25 |     "\n",
 26 |     "from sklearn.ensemble import RandomForestRegressor\n",
 27 |     "from sklearn.metrics import r2_score, roc_auc_score\n",
 28 |     "from sklearn import cross_validation\n",
 29 |     "from sklearn.model_selection import GridSearchCV\n",
 30 |     "from sklearn.svm import SVR "
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": 2,
 36 |    "metadata": {
 37 |     "collapsed": true
 38 |    },
 39 |    "outputs": [],
 40 |    "source": [
 41 |     "model_path = '../dataset/fetures/model.pkl'\n",
 42 |     "output_path = '../submit/Tianchi_power_predict_table.csv'"
 43 |    ]
 44 |   },
 45 |   {
 46 |    "cell_type": "code",
 47 |    "execution_count": 3,
 48 |    "metadata": {
 49 |     "collapsed": true
 50 |    },
 51 |    "outputs": [],
 52 |    "source": [
 53 |     "fetures = pd.read_csv('../dataset/fetures/test_feture.csv', parse_dates=['predict_date'])"
 54 |    ]
 55 |   },
 56 |   {
 57 |    "cell_type": "code",
 58 |    "execution_count": 4,
 59 |    "metadata": {
 60 |     "collapsed": true
 61 |    },
 62 |    "outputs": [],
 63 |    "source": [
 64 |     "model = pickle.load(open(model_path, 'rb'))"
 65 |    ]
 66 |   },
 67 |   {
 68 |    "cell_type": "code",
 69 |    "execution_count": 5,
 70 |    "metadata": {
 71 |     "collapsed": false
 72 |    },
 73 |    "outputs": [],
 74 |    "source": [
 75 |     "train_x = fetures.drop(['predict_date', 'predict_power_consumption'], axis=1)"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 6,
 81 |    "metadata": {
 82 |     "collapsed": false
 83 |    },
 84 |    "outputs": [],
 85 |    "source": [
 86 |     "fetures.predict_power_consumption = model.predict(train_x.fillna(0)).astype(np.int64)"
 87 |    ]
 88 |   },
 89 |   {
 90 |    "cell_type": "code",
 91 |    "execution_count": 7,
 92 |    "metadata": {
 93 |     "collapsed": false
 94 |    },
 95 |    "outputs": [],
 96 |    "source": [
 97 |     "fetures.predict_date = fetures.predict_date.map(lambda x: x.strftime('%Y%m%d'))"
 98 |    ]
 99 |   },
100 |   {
101 |    "cell_type": "code",
102 |    "execution_count": 8,
103 |    "metadata": {
104 |     "collapsed": false
105 |    },
106 |    "outputs": [
107 |     {
108 |      "data": {
109 |       "text/plain": [
110 |        "3754006.806451613"
111 |       ]
112 |      },
113 |      "execution_count": 8,
114 |      "metadata": {},
115 |      "output_type": "execute_result"
116 |     }
117 |    ],
118 |    "source": [
119 |     "fetures['predict_power_consumption'].mean()"
120 |    ]
121 |   },
122 |   {
123 |    "cell_type": "code",
124 |    "execution_count": 9,
125 |    "metadata": {
126 |     "collapsed": true
127 |    },
128 |    "outputs": [],
129 |    "source": [
130 |     "fetures.to_csv(output_path, index=False, columns=['predict_date', 'predict_power_consumption'])"
131 |    ]
132 |   }
133 |  ],
134 |  "metadata": {
135 |   "anaconda-cloud": {},
136 |   "hide_input": false,
137 |   "kernelspec": {
138 |    "display_name": "Python 3",
139 |    "language": "python",
140 |    "name": "python3"
141 |   },
142 |   "language_info": {
143 |    "codemirror_mode": {
144 |     "name": "ipython",
145 |     "version": 3
146 |    },
147 |    "file_extension": ".py",
148 |    "mimetype": "text/x-python",
149 |    "name": "python",
150 |    "nbconvert_exporter": "python",
151 |    "pygments_lexer": "ipython3",
152 |    "version": "3.5.1"
153 |   }
154 |  },
155 |  "nbformat": 4,
156 |  "nbformat_minor": 2
157 | }
158 | 


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/rules-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "from datetime import datetime, timedelta\n",
 12 |     "\n",
 13 |     "import numpy as np\n",
 14 |     "import pandas as pd\n",
 15 |     "import scipy as sp"
 16 |    ]
 17 |   },
 18 |   {
 19 |    "cell_type": "code",
 20 |    "execution_count": 2,
 21 |    "metadata": {
 22 |     "collapsed": true
 23 |    },
 24 |    "outputs": [],
 25 |    "source": [
 26 |     "indexs = pd.date_range('2016/10/1', '2016/10/31')"
 27 |    ]
 28 |   },
 29 |   {
 30 |    "cell_type": "code",
 31 |    "execution_count": 3,
 32 |    "metadata": {
 33 |     "collapsed": true
 34 |    },
 35 |    "outputs": [],
 36 |    "source": [
 37 |     "def gen_df(g):\n",
 38 |     "    df = pd.DataFrame(index=indexs, columns=['power_consumption'])\n",
 39 |     "    s = pd.Series(g.power_consumption, index=indexs)\n",
 40 |     "    df['power_consumption'] = s\n",
 41 |     "    df.fillna(df.power_consumption.mean(), inplace=True)\n",
 42 |     "    return df"
 43 |    ]
 44 |   },
 45 |   {
 46 |    "cell_type": "code",
 47 |    "execution_count": 4,
 48 |    "metadata": {
 49 |     "collapsed": true
 50 |    },
 51 |    "outputs": [],
 52 |    "source": [
 53 |     "train_path = '../dataset/train.csv'"
 54 |    ]
 55 |   },
 56 |   {
 57 |    "cell_type": "code",
 58 |    "execution_count": 5,
 59 |    "metadata": {
 60 |     "collapsed": true
 61 |    },
 62 |    "outputs": [],
 63 |    "source": [
 64 |     "train_df = pd.read_csv(train_path, parse_dates=['record_date'])"
 65 |    ]
 66 |   },
 67 |   {
 68 |    "cell_type": "code",
 69 |    "execution_count": 6,
 70 |    "metadata": {
 71 |     "collapsed": false
 72 |    },
 73 |    "outputs": [],
 74 |    "source": [
 75 |     "train_df['record_date'] = train_df.record_date + timedelta(days=31)"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 8,
 81 |    "metadata": {
 82 |     "collapsed": false
 83 |    },
 84 |    "outputs": [],
 85 |    "source": [
 86 |     "result_df = train_df.set_index('record_date').groupby(['user_id']).apply(gen_df)"
 87 |    ]
 88 |   },
 89 |   {
 90 |    "cell_type": "code",
 91 |    "execution_count": 9,
 92 |    "metadata": {
 93 |     "collapsed": false
 94 |    },
 95 |    "outputs": [],
 96 |    "source": [
 97 |     "submit_df = result_df.reset_index().groupby('level_1')[['power_consumption']].sum()"
 98 |    ]
 99 |   },
100 |   {
101 |    "cell_type": "code",
102 |    "execution_count": 10,
103 |    "metadata": {
104 |     "collapsed": false
105 |    },
106 |    "outputs": [],
107 |    "source": [
108 |     "submit_df['predict_date'] = submit_df.index.map(lambda x:x.strftime('%Y%m%d'))"
109 |    ]
110 |   },
111 |   {
112 |    "cell_type": "code",
113 |    "execution_count": 11,
114 |    "metadata": {
115 |     "collapsed": false
116 |    },
117 |    "outputs": [],
118 |    "source": [
119 |     "submit_df['predict_power_consumption'] = submit_df.power_consumption.astype(int)"
120 |    ]
121 |   },
122 |   {
123 |    "cell_type": "code",
124 |    "execution_count": 12,
125 |    "metadata": {
126 |     "collapsed": false
127 |    },
128 |    "outputs": [],
129 |    "source": [
130 |     "submit_df.to_csv('../Tianchi_power_predict_table.csv', columns=['predict_date', 'predict_power_consumption'], index=False)"
131 |    ]
132 |   }
133 |  ],
134 |  "metadata": {
135 |   "anaconda-cloud": {},
136 |   "hide_input": false,
137 |   "kernelspec": {
138 |    "display_name": "Python 3",
139 |    "language": "python",
140 |    "name": "python3"
141 |   },
142 |   "language_info": {
143 |    "codemirror_mode": {
144 |     "name": "ipython",
145 |     "version": 3
146 |    },
147 |    "file_extension": ".py",
148 |    "mimetype": "text/x-python",
149 |    "name": "python",
150 |    "nbconvert_exporter": "python",
151 |    "pygments_lexer": "ipython3",
152 |    "version": "3.5.1"
153 |   }
154 |  },
155 |  "nbformat": 4,
156 |  "nbformat_minor": 2
157 | }
158 | 


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/run-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": null,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "!date"
 12 |    ]
 13 |   },
 14 |   {
 15 |    "cell_type": "markdown",
 16 |    "metadata": {},
 17 |    "source": [
 18 |     "# 规则,前一个月平均值 "
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": null,
 24 |    "metadata": {
 25 |     "collapsed": true
 26 |    },
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "extract_ipy = 'trick_extract.ipynb'\n",
 30 |     "extract_py = 'trick_extract.py'"
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": null,
 36 |    "metadata": {
 37 |     "collapsed": false
 38 |    },
 39 |    "outputs": [],
 40 |    "source": [
 41 |     "# !jupyter nbconvert --to python $extract_ipy"
 42 |    ]
 43 |   },
 44 |   {
 45 |    "cell_type": "code",
 46 |    "execution_count": null,
 47 |    "metadata": {
 48 |     "collapsed": false
 49 |    },
 50 |    "outputs": [],
 51 |    "source": [
 52 |     "# !python $extract_py test"
 53 |    ]
 54 |   },
 55 |   {
 56 |    "cell_type": "markdown",
 57 |    "metadata": {},
 58 |    "source": [
 59 |     "## 切分数据"
 60 |    ]
 61 |   },
 62 |   {
 63 |    "cell_type": "code",
 64 |    "execution_count": null,
 65 |    "metadata": {
 66 |     "collapsed": false
 67 |    },
 68 |    "outputs": [],
 69 |    "source": [
 70 |     "!runipy split_samples.ipynb"
 71 |    ]
 72 |   },
 73 |   {
 74 |    "cell_type": "markdown",
 75 |    "metadata": {},
 76 |    "source": [
 77 |     "# 过滤清洗数据"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": null,
 83 |    "metadata": {
 84 |     "collapsed": false
 85 |    },
 86 |    "outputs": [],
 87 |    "source": [
 88 |     "!jupyter nbconvert --to python clean_data.ipynb"
 89 |    ]
 90 |   },
 91 |   {
 92 |    "cell_type": "code",
 93 |    "execution_count": null,
 94 |    "metadata": {
 95 |     "collapsed": false
 96 |    },
 97 |    "outputs": [],
 98 |    "source": [
 99 |     "!python clean_data.py test"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "code",
104 |    "execution_count": null,
105 |    "metadata": {
106 |     "collapsed": false
107 |    },
108 |    "outputs": [],
109 |    "source": [
110 |     "for i in range(1, 10):\n",
111 |     "    !python clean_data.py train $i"
112 |    ]
113 |   },
114 |   {
115 |    "cell_type": "markdown",
116 |    "metadata": {},
117 |    "source": [
118 |     "## 提取特征 "
119 |    ]
120 |   },
121 |   {
122 |    "cell_type": "code",
123 |    "execution_count": null,
124 |    "metadata": {
125 |     "collapsed": true
126 |    },
127 |    "outputs": [],
128 |    "source": [
129 |     "extract_ipy = 'extract_fetures.ipynb'\n",
130 |     "extract_py = 'extract_fetures.py'"
131 |    ]
132 |   },
133 |   {
134 |    "cell_type": "code",
135 |    "execution_count": null,
136 |    "metadata": {
137 |     "collapsed": false
138 |    },
139 |    "outputs": [],
140 |    "source": [
141 |     "!jupyter nbconvert --to python $extract_ipy"
142 |    ]
143 |   },
144 |   {
145 |    "cell_type": "code",
146 |    "execution_count": null,
147 |    "metadata": {
148 |     "collapsed": true
149 |    },
150 |    "outputs": [],
151 |    "source": [
152 |     "for i in range(1, 10):    \n",
153 |     "    !python $extract_py train $i"
154 |    ]
155 |   },
156 |   {
157 |    "cell_type": "code",
158 |    "execution_count": null,
159 |    "metadata": {
160 |     "collapsed": true
161 |    },
162 |    "outputs": [],
163 |    "source": [
164 |     "!python $extract_py test"
165 |    ]
166 |   },
167 |   {
168 |    "cell_type": "code",
169 |    "execution_count": null,
170 |    "metadata": {
171 |     "collapsed": true
172 |    },
173 |    "outputs": [],
174 |    "source": [
175 |     "!runipy combine_fetures.ipynb"
176 |    ]
177 |   },
178 |   {
179 |    "cell_type": "markdown",
180 |    "metadata": {},
181 |    "source": [
182 |     "## 训练模型 "
183 |    ]
184 |   },
185 |   {
186 |    "cell_type": "code",
187 |    "execution_count": null,
188 |    "metadata": {
189 |     "collapsed": false,
190 |     "scrolled": false
191 |    },
192 |    "outputs": [],
193 |    "source": [
194 |     "!runipy train_model.ipynb"
195 |    ]
196 |   },
197 |   {
198 |    "cell_type": "markdown",
199 |    "metadata": {},
200 |    "source": [
201 |     "## 模型预测 "
202 |    ]
203 |   },
204 |   {
205 |    "cell_type": "code",
206 |    "execution_count": null,
207 |    "metadata": {
208 |     "collapsed": false
209 |    },
210 |    "outputs": [],
211 |    "source": [
212 |     "!runipy predict_result.ipynb"
213 |    ]
214 |   },
215 |   {
216 |    "cell_type": "code",
217 |    "execution_count": null,
218 |    "metadata": {
219 |     "collapsed": false
220 |    },
221 |    "outputs": [],
222 |    "source": [
223 |     "!date"
224 |    ]
225 |   }
226 |  ],
227 |  "metadata": {
228 |   "anaconda-cloud": {},
229 |   "hide_input": false,
230 |   "kernelspec": {
231 |    "display_name": "Python 3",
232 |    "language": "python",
233 |    "name": "python3"
234 |   },
235 |   "language_info": {
236 |    "codemirror_mode": {
237 |     "name": "ipython",
238 |     "version": 3
239 |    },
240 |    "file_extension": ".py",
241 |    "mimetype": "text/x-python",
242 |    "name": "python",
243 |    "nbconvert_exporter": "python",
244 |    "pygments_lexer": "ipython3",
245 |    "version": "3.5.1"
246 |   }
247 |  },
248 |  "nbformat": 4,
249 |  "nbformat_minor": 2
250 | }
251 | 


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/split_samples-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 123,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "from os.path import exists, join\n",
 12 |     "from os import mkdir\n",
 13 |     "from datetime import datetime, timedelta, date\n",
 14 |     "from collections import namedtuple\n",
 15 |     "\n",
 16 |     "import numpy as np\n",
 17 |     "import pandas as pd\n",
 18 |     "import scipy as sp"
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": 124,
 24 |    "metadata": {
 25 |     "collapsed": true
 26 |    },
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "train_base_dir = '../dataset/fetures/'"
 30 |    ]
 31 |   },
 32 |   {
 33 |    "cell_type": "markdown",
 34 |    "metadata": {},
 35 |    "source": [
 36 |     "# 移动滑窗划分成9个predict样本 "
 37 |    ]
 38 |   },
 39 |   {
 40 |    "cell_type": "code",
 41 |    "execution_count": 125,
 42 |    "metadata": {
 43 |     "collapsed": false
 44 |    },
 45 |    "outputs": [],
 46 |    "source": [
 47 |     "df = pd.read_csv('../dataset/train.csv', parse_dates=['record_date'], index_col=['record_date'])"
 48 |    ]
 49 |   },
 50 |   {
 51 |    "cell_type": "code",
 52 |    "execution_count": 126,
 53 |    "metadata": {
 54 |     "collapsed": false
 55 |    },
 56 |    "outputs": [],
 57 |    "source": [
 58 |     "begin = df.index.max().date() + timedelta(1)"
 59 |    ]
 60 |   },
 61 |   {
 62 |    "cell_type": "markdown",
 63 |    "metadata": {},
 64 |    "source": [
 65 |     "## 训练样本划分"
 66 |    ]
 67 |   },
 68 |   {
 69 |    "cell_type": "code",
 70 |    "execution_count": 127,
 71 |    "metadata": {
 72 |     "collapsed": true
 73 |    },
 74 |    "outputs": [],
 75 |    "source": [
 76 |     "TimeSpan = namedtuple('TimeSpan', ['train_start', 'train_end', 'predict_start', 'predict_end'])"
 77 |    ]
 78 |   },
 79 |   {
 80 |    "cell_type": "code",
 81 |    "execution_count": 128,
 82 |    "metadata": {
 83 |     "collapsed": true
 84 |    },
 85 |    "outputs": [],
 86 |    "source": [
 87 |     "timespans = []"
 88 |    ]
 89 |   },
 90 |   {
 91 |    "cell_type": "code",
 92 |    "execution_count": 129,
 93 |    "metadata": {
 94 |     "collapsed": true
 95 |    },
 96 |    "outputs": [],
 97 |    "source": [
 98 |     "end = begin - timedelta(1)\n",
 99 |     "for i in range(9):\n",
100 |     "    p_end = date(begin.year, begin.month-i, begin.day)\n",
101 |     "    predict_end = p_end - timedelta(1)\n",
102 |     "    predict_start = date(begin.year, begin.month-i-1, begin.day)\n",
103 |     "    train_end = predict_start - timedelta(1)\n",
104 |     "    train_start = date(predict_start.year-1, predict_start.month, predict_start.day)\n",
105 |     "    span = TimeSpan(train_start, train_end, predict_start, predict_end)\n",
106 |     "#     print(span)\n",
107 |     "    timespans.append(span)"
108 |    ]
109 |   },
110 |   {
111 |    "cell_type": "markdown",
112 |    "metadata": {},
113 |    "source": [
114 |     "##### 创建文件架 "
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "code",
119 |    "execution_count": 130,
120 |    "metadata": {
121 |     "collapsed": false
122 |    },
123 |    "outputs": [],
124 |    "source": [
125 |     "for span in timespans:\n",
126 |     "    dir_n = str(span.predict_start.month)\n",
127 |     "    path = join(train_base_dir, dir_n)\n",
128 |     "    if not exists(path):\n",
129 |     "        mkdir(path)"
130 |    ]
131 |   },
132 |   {
133 |    "cell_type": "code",
134 |    "execution_count": 131,
135 |    "metadata": {
136 |     "collapsed": false
137 |    },
138 |    "outputs": [],
139 |    "source": [
140 |     "span = timespans[0]"
141 |    ]
142 |   },
143 |   {
144 |    "cell_type": "code",
145 |    "execution_count": 132,
146 |    "metadata": {
147 |     "collapsed": true
148 |    },
149 |    "outputs": [],
150 |    "source": [
151 |     "def create_train_sample(timespan):  \n",
152 |     "    train = df.loc[str(span.train_start):str(span.train_end)]    \n",
153 |     "    train_predict = df.loc[str(span.predict_start):str(span.predict_end)].reset_index()\n",
154 |     "    train_predict = train_predict.groupby('record_date')[['power_consumption']].sum()\n",
155 |     "    train_predict['predict_power_consumption'] = train_predict.power_consumption.astype(np.int64)\n",
156 |     "    train_predict.drop('power_consumption', axis=1, inplace=True)\n",
157 |     "    train_predict.index.name = 'predict_date'\n",
158 |     "    train_path = join(train_base_dir, str(span.predict_start.month), 'train.csv')\n",
159 |     "    train_predict_path = join(train_base_dir, str(span.predict_start.month), 'train_predict.csv')\n",
160 |     "    train.to_csv(train_path)\n",
161 |     "    train.info()\n",
162 |     "    train_predict.to_csv(train_predict_path)\n",
163 |     "    train_predict.info()"
164 |    ]
165 |   },
166 |   {
167 |    "cell_type": "code",
168 |    "execution_count": 133,
169 |    "metadata": {
170 |     "collapsed": false
171 |    },
172 |    "outputs": [
173 |     {
174 |      "name": "stdout",
175 |      "output_type": "stream",
176 |      "text": [
177 |       "<class 'pandas.core.frame.DataFrame'>\n",
178 |       "DatetimeIndex: 532156 entries, 2015-09-01 to 2016-08-31\n",
179 |       "Data columns (total 2 columns):\n",
180 |       "user_id              532156 non-null int64\n",
181 |       "power_consumption    532156 non-null int64\n",
182 |       "dtypes: int64(2)\n",
183 |       "memory usage: 12.2 MB\n",
184 |       "<class 'pandas.core.frame.DataFrame'>\n",
185 |       "DatetimeIndex: 30 entries, 2016-09-01 to 2016-09-30\n",
186 |       "Data columns (total 1 columns):\n",
187 |       "predict_power_consumption    30 non-null int64\n",
188 |       "dtypes: int64(1)\n",
189 |       "memory usage: 480.0 bytes\n",
190 |       "<class 'pandas.core.frame.DataFrame'>\n",
191 |       "DatetimeIndex: 532156 entries, 2015-08-01 to 2016-07-31\n",
192 |       "Data columns (total 2 columns):\n",
193 |       "user_id              532156 non-null int64\n",
194 |       "power_consumption    532156 non-null int64\n",
195 |       "dtypes: int64(2)\n",
196 |       "memory usage: 12.2 MB\n",
197 |       "<class 'pandas.core.frame.DataFrame'>\n",
198 |       "DatetimeIndex: 31 entries, 2016-08-01 to 2016-08-31\n",
199 |       "Data columns (total 1 columns):\n",
200 |       "predict_power_consumption    31 non-null int64\n",
201 |       "dtypes: int64(1)\n",
202 |       "memory usage: 496.0 bytes\n",
203 |       "<class 'pandas.core.frame.DataFrame'>\n",
204 |       "DatetimeIndex: 532153 entries, 2015-07-01 to 2016-06-30\n",
205 |       "Data columns (total 2 columns):\n",
206 |       "user_id              532153 non-null int64\n",
207 |       "power_consumption    532153 non-null int64\n",
208 |       "dtypes: int64(2)\n",
209 |       "memory usage: 12.2 MB\n",
210 |       "<class 'pandas.core.frame.DataFrame'>\n",
211 |       "DatetimeIndex: 31 entries, 2016-07-01 to 2016-07-31\n",
212 |       "Data columns (total 1 columns):\n",
213 |       "predict_power_consumption    31 non-null int64\n",
214 |       "dtypes: int64(1)\n",
215 |       "memory usage: 496.0 bytes\n",
216 |       "<class 'pandas.core.frame.DataFrame'>\n",
217 |       "DatetimeIndex: 532150 entries, 2015-06-01 to 2016-05-31\n",
218 |       "Data columns (total 2 columns):\n",
219 |       "user_id              532150 non-null int64\n",
220 |       "power_consumption    532150 non-null int64\n",
221 |       "dtypes: int64(2)\n",
222 |       "memory usage: 12.2 MB\n",
223 |       "<class 'pandas.core.frame.DataFrame'>\n",
224 |       "DatetimeIndex: 30 entries, 2016-06-01 to 2016-06-30\n",
225 |       "Data columns (total 1 columns):\n",
226 |       "predict_power_consumption    30 non-null int64\n",
227 |       "dtypes: int64(1)\n",
228 |       "memory usage: 480.0 bytes\n",
229 |       "<class 'pandas.core.frame.DataFrame'>\n",
230 |       "DatetimeIndex: 532150 entries, 2015-05-01 to 2016-04-30\n",
231 |       "Data columns (total 2 columns):\n",
232 |       "user_id              532150 non-null int64\n",
233 |       "power_consumption    532150 non-null int64\n",
234 |       "dtypes: int64(2)\n",
235 |       "memory usage: 12.2 MB\n",
236 |       "<class 'pandas.core.frame.DataFrame'>\n",
237 |       "DatetimeIndex: 31 entries, 2016-05-01 to 2016-05-31\n",
238 |       "Data columns (total 1 columns):\n",
239 |       "predict_power_consumption    31 non-null int64\n",
240 |       "dtypes: int64(1)\n",
241 |       "memory usage: 496.0 bytes\n",
242 |       "<class 'pandas.core.frame.DataFrame'>\n",
243 |       "DatetimeIndex: 532148 entries, 2015-04-01 to 2016-03-31\n",
244 |       "Data columns (total 2 columns):\n",
245 |       "user_id              532148 non-null int64\n",
246 |       "power_consumption    532148 non-null int64\n",
247 |       "dtypes: int64(2)\n",
248 |       "memory usage: 12.2 MB\n",
249 |       "<class 'pandas.core.frame.DataFrame'>\n",
250 |       "DatetimeIndex: 30 entries, 2016-04-01 to 2016-04-30\n",
251 |       "Data columns (total 1 columns):\n",
252 |       "predict_power_consumption    30 non-null int64\n",
253 |       "dtypes: int64(1)\n",
254 |       "memory usage: 480.0 bytes\n",
255 |       "<class 'pandas.core.frame.DataFrame'>\n",
256 |       "DatetimeIndex: 532147 entries, 2015-03-01 to 2016-02-29\n",
257 |       "Data columns (total 2 columns):\n",
258 |       "user_id              532147 non-null int64\n",
259 |       "power_consumption    532147 non-null int64\n",
260 |       "dtypes: int64(2)\n",
261 |       "memory usage: 12.2 MB\n",
262 |       "<class 'pandas.core.frame.DataFrame'>\n",
263 |       "DatetimeIndex: 31 entries, 2016-03-01 to 2016-03-31\n",
264 |       "Data columns (total 1 columns):\n",
265 |       "predict_power_consumption    31 non-null int64\n",
266 |       "dtypes: int64(1)\n",
267 |       "memory usage: 496.0 bytes\n",
268 |       "<class 'pandas.core.frame.DataFrame'>\n",
269 |       "DatetimeIndex: 530692 entries, 2015-02-01 to 2016-01-31\n",
270 |       "Data columns (total 2 columns):\n",
271 |       "user_id              530692 non-null int64\n",
272 |       "power_consumption    530692 non-null int64\n",
273 |       "dtypes: int64(2)\n",
274 |       "memory usage: 12.1 MB\n",
275 |       "<class 'pandas.core.frame.DataFrame'>\n",
276 |       "DatetimeIndex: 29 entries, 2016-02-01 to 2016-02-29\n",
277 |       "Data columns (total 1 columns):\n",
278 |       "predict_power_consumption    29 non-null int64\n",
279 |       "dtypes: int64(1)\n",
280 |       "memory usage: 464.0 bytes\n",
281 |       "<class 'pandas.core.frame.DataFrame'>\n",
282 |       "DatetimeIndex: 530693 entries, 2015-01-01 to 2015-12-31\n",
283 |       "Data columns (total 2 columns):\n",
284 |       "user_id              530693 non-null int64\n",
285 |       "power_consumption    530693 non-null int64\n",
286 |       "dtypes: int64(2)\n",
287 |       "memory usage: 12.1 MB\n",
288 |       "<class 'pandas.core.frame.DataFrame'>\n",
289 |       "DatetimeIndex: 31 entries, 2016-01-01 to 2016-01-31\n",
290 |       "Data columns (total 1 columns):\n",
291 |       "predict_power_consumption    31 non-null int64\n",
292 |       "dtypes: int64(1)\n",
293 |       "memory usage: 496.0 bytes\n"
294 |      ]
295 |     }
296 |    ],
297 |    "source": [
298 |     "for span in timespans:\n",
299 |     "    create_train_sample(span)"
300 |    ]
301 |   },
302 |   {
303 |    "cell_type": "code",
304 |    "execution_count": 138,
305 |    "metadata": {
306 |     "collapsed": true
307 |    },
308 |    "outputs": [],
309 |    "source": [
310 |     "test = df.loc['2015/10/1':'2016/9/30']"
311 |    ]
312 |   },
313 |   {
314 |    "cell_type": "code",
315 |    "execution_count": 139,
316 |    "metadata": {
317 |     "collapsed": false
318 |    },
319 |    "outputs": [],
320 |    "source": [
321 |     "index = pd.date_range('2016/10/1', '2016/10/31')\n",
322 |     "test_predict = pd.DataFrame(0, index=index, columns=['predict_power_consumption'])\n",
323 |     "test_predict.index.name = 'predict_date'"
324 |    ]
325 |   },
326 |   {
327 |    "cell_type": "code",
328 |    "execution_count": 140,
329 |    "metadata": {
330 |     "collapsed": false
331 |    },
332 |    "outputs": [
333 |     {
334 |      "name": "stdout",
335 |      "output_type": "stream",
336 |      "text": [
337 |       "<class 'pandas.core.frame.DataFrame'>\n",
338 |       "DatetimeIndex: 532159 entries, 2015-10-01 to 2016-09-28\n",
339 |       "Data columns (total 2 columns):\n",
340 |       "user_id              532159 non-null int64\n",
341 |       "power_consumption    532159 non-null int64\n",
342 |       "dtypes: int64(2)\n",
343 |       "memory usage: 12.2 MB\n"
344 |      ]
345 |     }
346 |    ],
347 |    "source": [
348 |     "test.to_csv('../dataset/fetures/test.csv')\n",
349 |     "test.info()"
350 |    ]
351 |   },
352 |   {
353 |    "cell_type": "code",
354 |    "execution_count": 142,
355 |    "metadata": {
356 |     "collapsed": false
357 |    },
358 |    "outputs": [
359 |     {
360 |      "name": "stdout",
361 |      "output_type": "stream",
362 |      "text": [
363 |       "<class 'pandas.core.frame.DataFrame'>\n",
364 |       "DatetimeIndex: 31 entries, 2016-10-01 to 2016-10-31\n",
365 |       "Freq: D\n",
366 |       "Data columns (total 1 columns):\n",
367 |       "predict_power_consumption    31 non-null int64\n",
368 |       "dtypes: int64(1)\n",
369 |       "memory usage: 496.0 bytes\n"
370 |      ]
371 |     }
372 |    ],
373 |    "source": [
374 |     "test_predict.to_csv('../dataset/fetures/test_predict.csv')\n",
375 |     "test_predict.info()"
376 |    ]
377 |   }
378 |  ],
379 |  "metadata": {
380 |   "anaconda-cloud": {},
381 |   "hide_input": false,
382 |   "kernelspec": {
383 |    "display_name": "Python 3",
384 |    "language": "python",
385 |    "name": "python3"
386 |   },
387 |   "language_info": {
388 |    "codemirror_mode": {
389 |     "name": "ipython",
390 |     "version": 3
391 |    },
392 |    "file_extension": ".py",
393 |    "mimetype": "text/x-python",
394 |    "name": "python",
395 |    "nbconvert_exporter": "python",
396 |    "pygments_lexer": "ipython3",
397 |    "version": "3.5.1"
398 |   }
399 |  },
400 |  "nbformat": 4,
401 |  "nbformat_minor": 2
402 | }
403 | 


--------------------------------------------------------------------------------
/code/.ipynb_checkpoints/train_model-checkpoint.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 35,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import pickle\n",
 12 |     "from datetime import datetime\n",
 13 |     "\n",
 14 |     "import numpy as np\n",
 15 |     "import scipy as sp\n",
 16 |     "import pandas as pd\n",
 17 |     "\n",
 18 |     "from sklearn.ensemble import RandomForestRegressor\n",
 19 |     "from sklearn.metrics import r2_score, roc_auc_score\n",
 20 |     "from sklearn import cross_validation\n",
 21 |     "from sklearn.model_selection import GridSearchCV\n",
 22 |     "from sklearn.svm import SVR\n",
 23 |     "import mlxtend"
 24 |    ]
 25 |   },
 26 |   {
 27 |    "cell_type": "code",
 28 |    "execution_count": 36,
 29 |    "metadata": {
 30 |     "collapsed": true
 31 |    },
 32 |    "outputs": [],
 33 |    "source": [
 34 |     "from mlxtend.regressor import StackingRegressor\n",
 35 |     "from mlxtend.data import boston_housing_data\n",
 36 |     "from sklearn.linear_model import LinearRegression\n",
 37 |     "from sklearn.linear_model import Ridge\n",
 38 |     "from sklearn.svm import SVR\n",
 39 |     "from sklearn.ensemble import ExtraTreesRegressor"
 40 |    ]
 41 |   },
 42 |   {
 43 |    "cell_type": "code",
 44 |    "execution_count": 37,
 45 |    "metadata": {
 46 |     "collapsed": true
 47 |    },
 48 |    "outputs": [],
 49 |    "source": [
 50 |     "def loss_score(predict, real):\n",
 51 |     "    f = (real - predict)/real\n",
 52 |     "    n = len(f)\n",
 53 |     "    f = f.replace([np.nan, -np.nan], 0)\n",
 54 |     "    score = 1 - np.abs(f).sum()/n\n",
 55 |     "    return score "
 56 |    ]
 57 |   },
 58 |   {
 59 |    "cell_type": "code",
 60 |    "execution_count": 38,
 61 |    "metadata": {
 62 |     "collapsed": true
 63 |    },
 64 |    "outputs": [],
 65 |    "source": [
 66 |     "model_path = '../dataset/fetures/model.pkl'"
 67 |    ]
 68 |   },
 69 |   {
 70 |    "cell_type": "code",
 71 |    "execution_count": 39,
 72 |    "metadata": {
 73 |     "collapsed": true
 74 |    },
 75 |    "outputs": [],
 76 |    "source": [
 77 |     "fetures = pd.read_csv('../dataset/fetures/train_feture.csv')"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": 40,
 83 |    "metadata": {
 84 |     "collapsed": false
 85 |    },
 86 |    "outputs": [],
 87 |    "source": [
 88 |     "train_x = fetures.drop(['predict_date', 'predict_power_consumption'], axis=1)"
 89 |    ]
 90 |   },
 91 |   {
 92 |    "cell_type": "code",
 93 |    "execution_count": 41,
 94 |    "metadata": {
 95 |     "collapsed": false
 96 |    },
 97 |    "outputs": [],
 98 |    "source": [
 99 |     "train_y = fetures['predict_power_consumption']"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "markdown",
104 |    "metadata": {},
105 |    "source": [
106 |     "###  使用 gridsearch 调参"
107 |    ]
108 |   },
109 |   {
110 |    "cell_type": "code",
111 |    "execution_count": 43,
112 |    "metadata": {
113 |     "collapsed": false
114 |    },
115 |    "outputs": [],
116 |    "source": [
117 |     "parameters = {\n",
118 |     "    'n_estimators': [10, 100, 200, 300, 500],\n",
119 |     "#     'criterion': ['mse', 'mae'],\n",
120 |     "    'max_features': [10, 50, 100, 200, 'sqrt']\n",
121 |     "}"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 44,
127 |    "metadata": {
128 |     "collapsed": true
129 |    },
130 |    "outputs": [],
131 |    "source": [
132 |     "# model = RandomForestRegressor(n_jobs=-1, oob_score=True)"
133 |    ]
134 |   },
135 |   {
136 |    "cell_type": "code",
137 |    "execution_count": 45,
138 |    "metadata": {
139 |     "collapsed": false
140 |    },
141 |    "outputs": [],
142 |    "source": [
143 |     "# cv = cross_validation.ShuffleSplit(train_x.shape[0], n_iter=8, test_size=0.1, random_state=0)\n",
144 |     "# clf = GridSearchCV(model, parameters, cv=cv)"
145 |    ]
146 |   },
147 |   {
148 |    "cell_type": "code",
149 |    "execution_count": 46,
150 |    "metadata": {
151 |     "collapsed": true
152 |    },
153 |    "outputs": [],
154 |    "source": [
155 |     "# clf.fit(train_x.fillna(0), train_y)"
156 |    ]
157 |   },
158 |   {
159 |    "cell_type": "code",
160 |    "execution_count": 47,
161 |    "metadata": {
162 |     "collapsed": false
163 |    },
164 |    "outputs": [],
165 |    "source": [
166 |     "# model = RandomForestRegressor(n_jobs=-1, oob_score=True, random_state=666, **clf.best_params_)"
167 |    ]
168 |   },
169 |   {
170 |    "cell_type": "code",
171 |    "execution_count": 48,
172 |    "metadata": {
173 |     "collapsed": false
174 |    },
175 |    "outputs": [],
176 |    "source": [
177 |     "# cv = cross_validation.ShuffleSplit(train_x.shape[0], n_iter=5, test_size=0.1, random_state=666)\n",
178 |     "# r2_score = cross_validation.cross_val_score(model, train_x.fillna(0), train_y, cv=cv)\n",
179 |     "# r2_score"
180 |    ]
181 |   },
182 |   {
183 |    "cell_type": "code",
184 |    "execution_count": 288,
185 |    "metadata": {
186 |     "collapsed": true
187 |    },
188 |    "outputs": [],
189 |    "source": [
190 |     "cv = cross_validation.ShuffleSplit(train_x.shape[0], n_iter=10, test_size=0.2, random_state=666)"
191 |    ]
192 |   },
193 |   {
194 |    "cell_type": "code",
195 |    "execution_count": 290,
196 |    "metadata": {
197 |     "collapsed": true
198 |    },
199 |    "outputs": [],
200 |    "source": [
201 |     "forest = RandomForestRegressor(\n",
202 |     "    n_jobs=-1,\n",
203 |     "#     criterion='mae',\n",
204 |     "    random_state=1,\n",
205 |     "    oob_score=True,\n",
206 |     "    n_estimators=100,\n",
207 |     "    max_features='auto')\n",
208 |     "fores = RandomForestRegressor(\n",
209 |     "    n_jobs=-1,\n",
210 |     "    random_state=2,\n",
211 |     "    oob_score=True,\n",
212 |     "    n_estimators=300,\n",
213 |     "    max_features=300)\n",
214 |     "extract = ExtraTreesRegressor(\n",
215 |     "    n_jobs=-1,\n",
216 |     "    random_state=3,\n",
217 |     "#     criterion='mae',\n",
218 |     "    bootstrap=True,\n",
219 |     "    oob_score=True,\n",
220 |     "    n_estimators=100,\n",
221 |     "    max_features='auto')\n",
222 |     "extrac = ExtraTreesRegressor(\n",
223 |     "    n_jobs=-1,\n",
224 |     "    random_state=4,\n",
225 |     "    bootstrap=True,\n",
226 |     "    oob_score=True,\n",
227 |     "    n_estimators=300,\n",
228 |     "    max_features=300)"
229 |    ]
230 |   },
231 |   {
232 |    "cell_type": "code",
233 |    "execution_count": 292,
234 |    "metadata": {
235 |     "collapsed": true
236 |    },
237 |    "outputs": [],
238 |    "source": [
239 |     "clfs = [forest, fores, extract, extrac]"
240 |    ]
241 |   },
242 |   {
243 |    "cell_type": "code",
244 |    "execution_count": 293,
245 |    "metadata": {
246 |     "collapsed": true
247 |    },
248 |    "outputs": [],
249 |    "source": [
250 |     "params = {\n",
251 |     "    'forest__n-estimators': [10, 50, 100],\n",
252 |     "    'forest__max-features': [10, 50, 100],\n",
253 |     "    'fores__n-estimators': [100, 200, 500],\n",
254 |     "    'fores__max-features': [100, 200, 300],\n",
255 |     "    'extract__n-estimators': [10, 50, 100],\n",
256 |     "    'extract__max-features': [10, 50, 100],\n",
257 |     "    'extrac__n-estimators': [100, 200, 500],\n",
258 |     "    'extrac__max-features': [100, 200, 300],\n",
259 |     "}"
260 |    ]
261 |   },
262 |   {
263 |    "cell_type": "code",
264 |    "execution_count": 294,
265 |    "metadata": {
266 |     "collapsed": false
267 |    },
268 |    "outputs": [],
269 |    "source": [
270 |     "meta = LinearRegression(n_jobs=-1)"
271 |    ]
272 |   },
273 |   {
274 |    "cell_type": "code",
275 |    "execution_count": 295,
276 |    "metadata": {
277 |     "collapsed": false
278 |    },
279 |    "outputs": [],
280 |    "source": [
281 |     "stacker = StackingRegressor(regressors=clfs, meta_regressor=meta, verbose=True)"
282 |    ]
283 |   },
284 |   {
285 |    "cell_type": "code",
286 |    "execution_count": 296,
287 |    "metadata": {
288 |     "collapsed": false
289 |    },
290 |    "outputs": [],
291 |    "source": [
292 |     "grid = GridSearchCV(estimator=stacker, param_grid=params, cv=cv)"
293 |    ]
294 |   },
295 |   {
296 |    "cell_type": "code",
297 |    "execution_count": 297,
298 |    "metadata": {
299 |     "collapsed": false
300 |    },
301 |    "outputs": [
302 |     {
303 |      "data": {
304 |       "text/plain": [
305 |        "{'bootstrap': True,\n",
306 |        " 'criterion': 'mse',\n",
307 |        " 'max_depth': None,\n",
308 |        " 'max_features': 'auto',\n",
309 |        " 'max_leaf_nodes': None,\n",
310 |        " 'min_impurity_split': 1e-07,\n",
311 |        " 'min_samples_leaf': 1,\n",
312 |        " 'min_samples_split': 2,\n",
313 |        " 'min_weight_fraction_leaf': 0.0,\n",
314 |        " 'n_estimators': 100,\n",
315 |        " 'n_jobs': -1,\n",
316 |        " 'oob_score': True,\n",
317 |        " 'random_state': 3,\n",
318 |        " 'verbose': 0,\n",
319 |        " 'warm_start': False}"
320 |       ]
321 |      },
322 |      "execution_count": 297,
323 |      "metadata": {},
324 |      "output_type": "execute_result"
325 |     }
326 |    ],
327 |    "source": [
328 |     "extract.get_params()"
329 |    ]
330 |   },
331 |   {
332 |    "cell_type": "code",
333 |    "execution_count": 298,
334 |    "metadata": {
335 |     "collapsed": true
336 |    },
337 |    "outputs": [],
338 |    "source": [
339 |     "# grid.fit(train_x.fillna(0), train_y)"
340 |    ]
341 |   },
342 |   {
343 |    "cell_type": "code",
344 |    "execution_count": 300,
345 |    "metadata": {
346 |     "collapsed": true
347 |    },
348 |    "outputs": [
349 |     {
350 |      "name": "stdout",
351 |      "output_type": "stream",
352 |      "text": [
353 |       "Fitting 4 regressors...\n",
354 |       "Fitting regressor1: randomforestregressor (1/4)\n",
355 |       "Fitting 4 regressors...\n",
356 |       "Fitting regressor1: randomforestregressor (1/4)\n",
357 |       "Fitting 4 regressors...\n",
358 |       "Fitting regressor1: randomforestregressor (1/4)\n",
359 |       "Fitting 4 regressors...\n",
360 |       "Fitting regressor1: randomforestregressor (1/4)\n",
361 |       "Fitting regressor2: randomforestregressor (2/4)\n",
362 |       "Fitting regressor2: randomforestregressor (2/4)\n",
363 |       "Fitting regressor2: randomforestregressor (2/4)\n",
364 |       "Fitting regressor2: randomforestregressor (2/4)\n",
365 |       "Fitting regressor3: extratreesregressor (3/4)\n",
366 |       "Fitting regressor3: extratreesregressor (3/4)\n",
367 |       "Fitting regressor3: extratreesregressor (3/4)\n",
368 |       "Fitting regressor3: extratreesregressor (3/4)\n",
369 |       "Fitting regressor4: extratreesregressor (4/4)\n",
370 |       "Fitting regressor4: extratreesregressor (4/4)\n",
371 |       "Fitting regressor4: extratreesregressor (4/4)\n",
372 |       "Fitting regressor4: extratreesregressor (4/4)\n",
373 |       "Fitting 4 regressors...\n",
374 |       "Fitting regressor1: randomforestregressor (1/4)\n",
375 |       "Fitting 4 regressors...\n",
376 |       "Fitting regressor1: randomforestregressor (1/4)\n",
377 |       "Fitting 4 regressors...\n",
378 |       "Fitting regressor1: randomforestregressor (1/4)\n",
379 |       "Fitting 4 regressors...\n",
380 |       "Fitting regressor1: randomforestregressor (1/4)\n",
381 |       "Fitting regressor2: randomforestregressor (2/4)\n",
382 |       "Fitting regressor2: randomforestregressor (2/4)\n",
383 |       "Fitting regressor2: randomforestregressor (2/4)\n",
384 |       "Fitting regressor2: randomforestregressor (2/4)\n",
385 |       "Fitting regressor3: extratreesregressor (3/4)\n",
386 |       "Fitting regressor3: extratreesregressor (3/4)\n",
387 |       "Fitting regressor3: extratreesregressor (3/4)\n",
388 |       "Fitting regressor3: extratreesregressor (3/4)\n",
389 |       "Fitting regressor4: extratreesregressor (4/4)\n",
390 |       "Fitting regressor4: extratreesregressor (4/4)\n",
391 |       "Fitting regressor4: extratreesregressor (4/4)\n",
392 |       "Fitting regressor4: extratreesregressor (4/4)\n",
393 |       "Fitting 4 regressors...\n",
394 |       "Fitting regressor1: randomforestregressor (1/4)\n",
395 |       "Fitting 4 regressors...\n",
396 |       "Fitting regressor1: randomforestregressor (1/4)\n",
397 |       "Fitting regressor2: randomforestregressor (2/4)\n",
398 |       "Fitting regressor2: randomforestregressor (2/4)\n",
399 |       "Fitting regressor3: extratreesregressor (3/4)\n",
400 |       "Fitting regressor3: extratreesregressor (3/4)\n",
401 |       "Fitting regressor4: extratreesregressor (4/4)\n",
402 |       "Fitting regressor4: extratreesregressor (4/4)\n"
403 |      ]
404 |     },
405 |     {
406 |      "name": "stderr",
407 |      "output_type": "stream",
408 |      "text": [
409 |       "[Parallel(n_jobs=-1)]: Done  10 out of  10 | elapsed:  2.1min finished\n"
410 |      ]
411 |     }
412 |    ],
413 |    "source": [
414 |     "r2_score = cross_validation.cross_val_score(stacker, train_x, train_y, cv=cv, verbose=True, n_jobs=-1)"
415 |    ]
416 |   },
417 |   {
418 |    "cell_type": "code",
419 |    "execution_count": 301,
420 |    "metadata": {
421 |     "collapsed": true
422 |    },
423 |    "outputs": [],
424 |    "source": [
425 |     "# grid.fit(train_x.fillna(0), train_y)"
426 |    ]
427 |   },
428 |   {
429 |    "cell_type": "code",
430 |    "execution_count": 302,
431 |    "metadata": {
432 |     "collapsed": false
433 |    },
434 |    "outputs": [
435 |     {
436 |      "data": {
437 |       "text/plain": [
438 |        "(0.78898896165565757, 0.86008654041103993, 0.68182410751042843)"
439 |       ]
440 |      },
441 |      "execution_count": 302,
442 |      "metadata": {},
443 |      "output_type": "execute_result"
444 |     }
445 |    ],
446 |    "source": [
447 |     "r2_score.mean(), r2_score.max(), r2_score.min()"
448 |    ]
449 |   },
450 |   {
451 |    "cell_type": "code",
452 |    "execution_count": 303,
453 |    "metadata": {
454 |     "collapsed": false
455 |    },
456 |    "outputs": [
457 |     {
458 |      "name": "stdout",
459 |      "output_type": "stream",
460 |      "text": [
461 |       "Fitting 4 regressors...\n",
462 |       "Fitting regressor1: randomforestregressor (1/4)\n",
463 |       "Fitting regressor2: randomforestregressor (2/4)\n",
464 |       "Fitting regressor3: extratreesregressor (3/4)\n",
465 |       "Fitting regressor4: extratreesregressor (4/4)\n"
466 |      ]
467 |     },
468 |     {
469 |      "data": {
470 |       "text/plain": [
471 |        "StackingRegressor(meta_regressor=LinearRegression(copy_X=True, fit_intercept=True, n_jobs=-1, normalize=False),\n",
472 |        "         regressors=[RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
473 |        "           max_features='auto', max_leaf_nodes=None,\n",
474 |        "           min_impurity_split=1e-07, min_samples_leaf=1,\n",
475 |        "           min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
476 |        "           n_estimators=100, n_jobs=-1, oob_s...n_estimators=300, n_jobs=-1,\n",
477 |        "          oob_score=True, random_state=4, verbose=0, warm_start=False)],\n",
478 |        "         verbose=True)"
479 |       ]
480 |      },
481 |      "execution_count": 303,
482 |      "metadata": {},
483 |      "output_type": "execute_result"
484 |     }
485 |    ],
486 |    "source": [
487 |     "stacker.fit(train_x.fillna(0), train_y)"
488 |    ]
489 |   },
490 |   {
491 |    "cell_type": "code",
492 |    "execution_count": 304,
493 |    "metadata": {
494 |     "collapsed": false
495 |    },
496 |    "outputs": [],
497 |    "source": [
498 |     "r2_all_scorr = stacker.score(train_x.fillna(0), train_y)"
499 |    ]
500 |   },
501 |   {
502 |    "cell_type": "code",
503 |    "execution_count": null,
504 |    "metadata": {
505 |     "collapsed": true
506 |    },
507 |    "outputs": [],
508 |    "source": [
509 |     "# importance_df = pd.DataFrame(model.feature_importances_, index=train_x.columns)\n",
510 |     "\n",
511 |     "# importance_df.sort_values(0, ascending=False, inplace=True)\n",
512 |     "\n",
513 |     "# importance_top20 = ' , '.join(['{}:{} '.format(x, importance_df.loc[x].values[0]) for x in importance_df.index][:20])\n",
514 |     "# importance_top20"
515 |    ]
516 |   },
517 |   {
518 |    "cell_type": "code",
519 |    "execution_count": 305,
520 |    "metadata": {
521 |     "collapsed": false
522 |    },
523 |    "outputs": [
524 |     {
525 |      "data": {
526 |       "text/plain": [
527 |        "0.9876500688451734"
528 |       ]
529 |      },
530 |      "execution_count": 305,
531 |      "metadata": {},
532 |      "output_type": "execute_result"
533 |     }
534 |    ],
535 |    "source": [
536 |     "predict_result = stacker.predict(train_x.fillna(0))\n",
537 |     "loss_s  = loss_score(predict_result, train_y)\n",
538 |     "loss_s"
539 |    ]
540 |   },
541 |   {
542 |    "cell_type": "code",
543 |    "execution_count": 306,
544 |    "metadata": {
545 |     "collapsed": false
546 |    },
547 |    "outputs": [
548 |     {
549 |      "data": {
550 |       "text/plain": [
551 |        "(0.9876500688451734,\n",
552 |        " 0.98078256144556675,\n",
553 |        " 0.78898896165565757,\n",
554 |        " 0.86008654041103993,\n",
555 |        " 0.68182410751042843)"
556 |       ]
557 |      },
558 |      "execution_count": 306,
559 |      "metadata": {},
560 |      "output_type": "execute_result"
561 |     }
562 |    ],
563 |    "source": [
564 |     "result = loss_s, r2_all_scorr, r2_score.mean(), r2_score.max(), r2_score.min()\n",
565 |     "result"
566 |    ]
567 |   },
568 |   {
569 |    "cell_type": "code",
570 |    "execution_count": 307,
571 |    "metadata": {
572 |     "collapsed": false
573 |    },
574 |    "outputs": [
575 |     {
576 |      "name": "stdout",
577 |      "output_type": "stream",
578 |      "text": [
579 |       "2017-06-13 07:55:11.228802 result: (0.9876500688451734, 0.98078256144556675, 0.78898896165565757, 0.86008654041103993, 0.68182410751042843)\n"
580 |      ]
581 |     }
582 |    ],
583 |    "source": [
584 |     "print(datetime.now(), 'result:', result)\n",
585 |     "print(\n",
586 |     "    datetime.now(),\n",
587 |     "    'result:', result,\n",
588 |     "    file=open('result.txt', 'a+'),\n",
589 |     "    sep=' ',\n",
590 |     "    end='\\n')"
591 |    ]
592 |   },
593 |   {
594 |    "cell_type": "code",
595 |    "execution_count": 308,
596 |    "metadata": {
597 |     "collapsed": false
598 |    },
599 |    "outputs": [],
600 |    "source": [
601 |     "pickle.dump(stacker, open(model_path, 'wb'))"
602 |    ]
603 |   }
604 |  ],
605 |  "metadata": {
606 |   "anaconda-cloud": {},
607 |   "hide_input": false,
608 |   "kernelspec": {
609 |    "display_name": "Python 3",
610 |    "language": "python",
611 |    "name": "python3"
612 |   },
613 |   "language_info": {
614 |    "codemirror_mode": {
615 |     "name": "ipython",
616 |     "version": 3
617 |    },
618 |    "file_extension": ".py",
619 |    "mimetype": "text/x-python",
620 |    "name": "python",
621 |    "nbconvert_exporter": "python",
622 |    "pygments_lexer": "ipython3",
623 |    "version": "3.5.1"
624 |   }
625 |  },
626 |  "nbformat": 4,
627 |  "nbformat_minor": 2
628 | }
629 | 


--------------------------------------------------------------------------------
/code/clean_data.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import sys\n",
 12 |     "from datetime import datetime, timedelta, date\n",
 13 |     "\n",
 14 |     "import numpy as np\n",
 15 |     "import pandas as pd\n",
 16 |     "import scipy as sp\n",
 17 |     "\n",
 18 |     "from sklearn.ensemble import IsolationForest"
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": 2,
 24 |    "metadata": {
 25 |     "collapsed": false
 26 |    },
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "if sys.argv[1] == 'test':\n",
 30 |     "    is_train = False\n",
 31 |     "    train_path = '../dataset/fetures/test.csv'\n",
 32 |     "else:\n",
 33 |     "    is_train = True\n",
 34 |     "    if sys.argv[2].endswith('.json'):\n",
 35 |     "        month = '9'\n",
 36 |     "    else:\n",
 37 |     "        month = sys.argv[2]\n",
 38 |     "    train_path = '../dataset/fetures/{}/train.csv'.format(month)"
 39 |    ]
 40 |   },
 41 |   {
 42 |    "cell_type": "code",
 43 |    "execution_count": 3,
 44 |    "metadata": {
 45 |     "collapsed": false
 46 |    },
 47 |    "outputs": [],
 48 |    "source": [
 49 |     "train = pd.read_csv(train_path, parse_dates=['record_date'])"
 50 |    ]
 51 |   },
 52 |   {
 53 |    "cell_type": "code",
 54 |    "execution_count": 4,
 55 |    "metadata": {
 56 |     "collapsed": false
 57 |    },
 58 |    "outputs": [
 59 |     {
 60 |      "name": "stdout",
 61 |      "output_type": "stream",
 62 |      "text": [
 63 |       "<class 'pandas.core.frame.DataFrame'>\n",
 64 |       "RangeIndex: 532156 entries, 0 to 532155\n",
 65 |       "Data columns (total 3 columns):\n",
 66 |       "record_date          532156 non-null datetime64[ns]\n",
 67 |       "user_id              532156 non-null int64\n",
 68 |       "power_consumption    532156 non-null int64\n",
 69 |       "dtypes: datetime64[ns](1), int64(2)\n",
 70 |       "memory usage: 12.2 MB\n"
 71 |      ]
 72 |     }
 73 |    ],
 74 |    "source": [
 75 |     "train.info()"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 5,
 81 |    "metadata": {
 82 |     "collapsed": true
 83 |    },
 84 |    "outputs": [],
 85 |    "source": [
 86 |     "threshold = 0.95"
 87 |    ]
 88 |   },
 89 |   {
 90 |    "cell_type": "code",
 91 |    "execution_count": 6,
 92 |    "metadata": {
 93 |     "collapsed": true
 94 |    },
 95 |    "outputs": [],
 96 |    "source": [
 97 |     "def clean_quantile(df):\n",
 98 |     "    up = df.power_consumption.quantile(threshold)\n",
 99 |     "    down = df.power_consumption.quantile(1-threshold)\n",
100 |     "    mean = df.power_consumption.mean()\n",
101 |     "    print(df['user_id'].iloc[0], 'up:', up, 'down:', down, 'mean:', mean, end='\\n', file=open('../dataset/clean.txt', 'a+'))\n",
102 |     "    df.loc[(df.power_consumption>=up)|(df.power_consumption<=down), 'power_consumption'] = df.power_consumption.mean()\n",
103 |     "    return df"
104 |    ]
105 |   },
106 |   {
107 |    "cell_type": "code",
108 |    "execution_count": 7,
109 |    "metadata": {
110 |     "collapsed": false
111 |    },
112 |    "outputs": [],
113 |    "source": [
114 |     "train = train.groupby('user_id', as_index=True).apply(clean_quantile)"
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "markdown",
119 |    "metadata": {},
120 |    "source": [
121 |     "# 去除这些公司后结果变差,可能是因为未来也会造成影响"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 9,
127 |    "metadata": {
128 |     "collapsed": true
129 |    },
130 |    "outputs": [],
131 |    "source": [
132 |     "def clean_all_zero(df):\n",
133 |     "    ndf = df.loc[df.power_consumption!=1]\n",
134 |     "    if not ndf.empty:\n",
135 |     "        return df\n",
136 |     "    else:\n",
137 |     "        print(df['user_id'].iloc[0])\n",
138 |     "        return ndf\n"
139 |    ]
140 |   },
141 |   {
142 |    "cell_type": "code",
143 |    "execution_count": 10,
144 |    "metadata": {
145 |     "collapsed": true
146 |    },
147 |    "outputs": [
148 |     {
149 |      "name": "stdout",
150 |      "output_type": "stream",
151 |      "text": [
152 |       "259\n",
153 |       "413\n",
154 |       "418\n",
155 |       "462\n",
156 |       "469\n",
157 |       "522\n",
158 |       "533\n",
159 |       "549\n",
160 |       "550\n",
161 |       "551\n",
162 |       "553\n",
163 |       "557\n",
164 |       "570\n",
165 |       "586\n",
166 |       "588\n",
167 |       "589\n",
168 |       "598\n",
169 |       "606\n",
170 |       "609\n",
171 |       "617\n",
172 |       "622\n",
173 |       "633\n",
174 |       "634\n",
175 |       "637\n",
176 |       "638\n",
177 |       "639\n",
178 |       "645\n",
179 |       "646\n",
180 |       "647\n",
181 |       "650\n",
182 |       "651\n",
183 |       "652\n",
184 |       "653\n",
185 |       "654\n",
186 |       "655\n",
187 |       "656\n",
188 |       "657\n",
189 |       "658\n",
190 |       "661\n",
191 |       "663\n",
192 |       "664\n",
193 |       "665\n",
194 |       "666\n",
195 |       "667\n",
196 |       "668\n",
197 |       "669\n",
198 |       "670\n",
199 |       "671\n",
200 |       "672\n",
201 |       "673\n",
202 |       "674\n",
203 |       "675\n",
204 |       "676\n",
205 |       "677\n",
206 |       "678\n",
207 |       "679\n",
208 |       "680\n",
209 |       "681\n",
210 |       "682\n",
211 |       "683\n",
212 |       "684\n",
213 |       "685\n",
214 |       "686\n",
215 |       "687\n",
216 |       "688\n",
217 |       "689\n",
218 |       "690\n",
219 |       "691\n",
220 |       "692\n",
221 |       "693\n",
222 |       "694\n",
223 |       "695\n",
224 |       "696\n",
225 |       "709\n",
226 |       "891\n",
227 |       "950\n",
228 |       "1044\n",
229 |       "1045\n",
230 |       "1111\n",
231 |       "1129\n",
232 |       "1233\n",
233 |       "1234\n",
234 |       "1261\n",
235 |       "1280\n",
236 |       "1413\n",
237 |       "<class 'pandas.core.frame.DataFrame'>\n",
238 |       "RangeIndex: 501046 entries, 0 to 501045\n",
239 |       "Data columns (total 3 columns):\n",
240 |       "record_date          501046 non-null datetime64[ns]\n",
241 |       "user_id              501046 non-null int64\n",
242 |       "power_consumption    501046 non-null float64\n",
243 |       "dtypes: datetime64[ns](1), float64(1), int64(1)\n",
244 |       "memory usage: 11.5 MB\n"
245 |      ]
246 |     }
247 |    ],
248 |    "source": [
249 |     "# train = train.groupby('user_id').apply(clean_all_zero).reset_index(drop=True)\n",
250 |     "# train.info()"
251 |    ]
252 |   },
253 |   {
254 |    "cell_type": "code",
255 |    "execution_count": null,
256 |    "metadata": {
257 |     "collapsed": true
258 |    },
259 |    "outputs": [],
260 |    "source": [
261 |     "# test = test.groupby('user_id').apply(clean_all_zero).reset_index(drop=True)\n",
262 |     "# test.info()"
263 |    ]
264 |   },
265 |   {
266 |    "cell_type": "markdown",
267 |    "metadata": {},
268 |    "source": [
269 |     "# 添加公司用电量平滑处理"
270 |    ]
271 |   },
272 |   {
273 |    "cell_type": "code",
274 |    "execution_count": 11,
275 |    "metadata": {
276 |     "collapsed": true
277 |    },
278 |    "outputs": [],
279 |    "source": [
280 |     "def rolling_power_consumption(df):\n",
281 |     "    ndf = df.set_index('record_date')\n",
282 |     "    ndf = ndf.rolling(2).mean()\n",
283 |     "    return ndf.reset_index()"
284 |    ]
285 |   },
286 |   {
287 |    "cell_type": "code",
288 |    "execution_count": 12,
289 |    "metadata": {
290 |     "collapsed": false
291 |    },
292 |    "outputs": [],
293 |    "source": [
294 |     "train = train.groupby('user_id').apply(rolling_power_consumption).dropna(subset=['user_id']).reset_index(drop=True)"
295 |    ]
296 |   },
297 |   {
298 |    "cell_type": "code",
299 |    "execution_count": 13,
300 |    "metadata": {
301 |     "collapsed": false
302 |    },
303 |    "outputs": [
304 |     {
305 |      "name": "stdout",
306 |      "output_type": "stream",
307 |      "text": [
308 |       "<class 'pandas.core.frame.DataFrame'>\n",
309 |       "RangeIndex: 499677 entries, 0 to 499676\n",
310 |       "Data columns (total 3 columns):\n",
311 |       "record_date          499677 non-null datetime64[ns]\n",
312 |       "user_id              499677 non-null float64\n",
313 |       "power_consumption    499677 non-null float64\n",
314 |       "dtypes: datetime64[ns](1), float64(2)\n",
315 |       "memory usage: 11.4 MB\n"
316 |      ]
317 |     }
318 |    ],
319 |    "source": [
320 |     "train.to_csv(train_path, index=False)\n",
321 |     "train.info()"
322 |    ]
323 |   }
324 |  ],
325 |  "metadata": {
326 |   "anaconda-cloud": {},
327 |   "hide_input": false,
328 |   "kernelspec": {
329 |    "display_name": "Python 3",
330 |    "language": "python",
331 |    "name": "python3"
332 |   },
333 |   "language_info": {
334 |    "codemirror_mode": {
335 |     "name": "ipython",
336 |     "version": 3
337 |    },
338 |    "file_extension": ".py",
339 |    "mimetype": "text/x-python",
340 |    "name": "python",
341 |    "nbconvert_exporter": "python",
342 |    "pygments_lexer": "ipython3",
343 |    "version": "3.5.1"
344 |   }
345 |  },
346 |  "nbformat": 4,
347 |  "nbformat_minor": 2
348 | }
349 | 


--------------------------------------------------------------------------------
/code/clean_data.py:
--------------------------------------------------------------------------------
  1 | 
  2 | # coding: utf-8
  3 | 
  4 | # In[1]:
  5 | 
  6 | import sys
  7 | from datetime import datetime, timedelta, date
  8 | 
  9 | import numpy as np
 10 | import pandas as pd
 11 | import scipy as sp
 12 | 
 13 | from sklearn.ensemble import IsolationForest
 14 | 
 15 | 
 16 | # In[2]:
 17 | 
 18 | if sys.argv[1] == 'test':
 19 |     is_train = False
 20 |     train_path = '../dataset/fetures/test.csv'
 21 | else:
 22 |     is_train = True
 23 |     if sys.argv[2].endswith('.json'):
 24 |         month = '9'
 25 |     else:
 26 |         month = sys.argv[2]
 27 |     train_path = '../dataset/fetures/{}/train.csv'.format(month)
 28 | 
 29 | 
 30 | # In[3]:
 31 | 
 32 | train = pd.read_csv(train_path, parse_dates=['record_date'])
 33 | 
 34 | 
 35 | # In[4]:
 36 | 
 37 | train.info()
 38 | 
 39 | 
 40 | # In[5]:
 41 | 
 42 | threshold = 0.95
 43 | 
 44 | 
 45 | # In[6]:
 46 | 
 47 | def clean_quantile(df):
 48 |     up = df.power_consumption.quantile(threshold)
 49 |     down = df.power_consumption.quantile(1-threshold)
 50 |     mean = df.power_consumption.mean()
 51 |     print(df['user_id'].iloc[0], 'up:', up, 'down:', down, 'mean:', mean, end='\n', file=open('../dataset/clean.txt', 'a+'))
 52 |     df.loc[(df.power_consumption>=up)|(df.power_consumption<=down), 'power_consumption'] = df.power_consumption.mean()
 53 |     return df
 54 | 
 55 | 
 56 | # In[7]:
 57 | 
 58 | train = train.groupby('user_id', as_index=True).apply(clean_quantile)
 59 | 
 60 | 
 61 | # # 去除这些公司后结果变差,可能是因为未来也会造成影响
 62 | 
 63 | # In[9]:
 64 | 
 65 | def clean_all_zero(df):
 66 |     ndf = df.loc[df.power_consumption!=1]
 67 |     if not ndf.empty:
 68 |         return df
 69 |     else:
 70 |         print(df['user_id'].iloc[0])
 71 |         return ndf
 72 | 
 73 | 
 74 | # In[10]:
 75 | 
 76 | # train = train.groupby('user_id').apply(clean_all_zero).reset_index(drop=True)
 77 | # train.info()
 78 | 
 79 | 
 80 | # In[ ]:
 81 | 
 82 | # test = test.groupby('user_id').apply(clean_all_zero).reset_index(drop=True)
 83 | # test.info()
 84 | 
 85 | 
 86 | # # 添加公司用电量平滑处理
 87 | 
 88 | # In[11]:
 89 | 
 90 | def rolling_power_consumption(df):
 91 |     ndf = df.set_index('record_date')
 92 |     ndf = ndf.rolling(2).mean()
 93 |     return ndf.reset_index()
 94 | 
 95 | 
 96 | # In[12]:
 97 | 
 98 | train = train.groupby('user_id').apply(rolling_power_consumption).dropna(subset=['user_id']).reset_index(drop=True)
 99 | 
100 | 
101 | # In[13]:
102 | 
103 | train.to_csv(train_path, index=False)
104 | train.info()
105 | 
106 | 


--------------------------------------------------------------------------------
/code/extract_fetures.py:
--------------------------------------------------------------------------------
  1 | 
  2 | # coding: utf-8
  3 | 
  4 | # In[1]:
  5 | 
  6 | import sys
  7 | from datetime import datetime, timedelta, date
  8 | 
  9 | import numpy as np
 10 | import pandas as pd
 11 | import scipy as sp
 12 | 
 13 | 
 14 | # In[2]:
 15 | 
 16 | def loss_score(predict, real):
 17 |     f = (real - predict)/real
 18 |     n = len(f)
 19 |     f = f.replace([np.nan, -np.nan], 0)
 20 |     score = 1 - np.abs(f).sum()/n
 21 |     return score 
 22 | 
 23 | 
 24 | # In[3]:
 25 | 
 26 | if sys.argv[1] == 'test':
 27 |     is_train = False
 28 |     train_path = '../dataset/fetures/test.csv'
 29 |     predict_path = '../dataset/fetures/test_predict.csv'
 30 |     feture_path = '../dataset/fetures/test_feture.csv'
 31 |     
 32 | else:
 33 |     is_train = True
 34 |     if sys.argv[2].endswith('.json'):
 35 |         month = '9'
 36 |     else:
 37 |         month = sys.argv[2]
 38 |     train_path = '../dataset/fetures/{}/train.csv'.format(month)
 39 |     predict_path = '../dataset/fetures/{}/train_predict.csv'.format(month)
 40 |     feture_path = '../dataset/fetures/{}/train_feture.csv'.format(month)
 41 |     
 42 | 
 43 | 
 44 | # In[4]:
 45 | 
 46 | train = pd.read_csv(train_path, parse_dates=['record_date'])
 47 | predict = pd.read_csv(predict_path, parse_dates=['predict_date'])
 48 | 
 49 | 
 50 | # In[5]:
 51 | 
 52 | end_date = train.record_date.max().date()
 53 | 
 54 | 
 55 | # In[6]:
 56 | 
 57 | def create_timespan(end_date, m_span):
 58 |     predict_start = end_date + timedelta(1)
 59 |     n_month = predict_start.month - m_span
 60 |     if n_month < 1:
 61 |         n_month = 12 + n_month
 62 |         n_year = predict_start.year - 1
 63 |     else:
 64 |         n_year = predict_start.year
 65 |     n_date = date(n_year, n_month, predict_start.day)
 66 |     return (predict_start - n_date).days - 1
 67 | 
 68 | 
 69 | # In[7]:
 70 | 
 71 | # create_timespan(end_date, 13)
 72 | 
 73 | 
 74 | # In[8]:
 75 | 
 76 | time_spans = [30, 60, 90, 180, 366]
 77 | # time_spans = [30, 60, 90, 150, 180, 210, 270, 300, 366] # over fit
 78 | 
 79 | 
 80 | # In[9]:
 81 | 
 82 | # for span in time_month_spans:
 83 | #     time_spans.append( create_timespan(end_date, span) )
 84 | 
 85 | 
 86 | # In[10]:
 87 | 
 88 | print(time_spans)
 89 | 
 90 | 
 91 | # In[11]:
 92 | 
 93 | used_data_fetures = [
 94 |     'dayofweek', 'dayofyear', 'days_in_month', 'quarter', 'week', 'weekofyear',
 95 |     'month', 'year'
 96 | ]
 97 | 
 98 | 
 99 | # In[12]:
100 | 
101 | def get_train_df(date_span, train):
102 |     begin_date = end_date - timedelta(date_span)
103 |     train_df = train.set_index(['record_date']).loc[str(begin_date):str(end_date)].reset_index()
104 |     return train_df
105 | 
106 | 
107 | # In[13]:
108 | 
109 | def add_feture_in_date(used_feture, predict, train_df):
110 |     for f in used_data_fetures:
111 |         predict[f] = getattr(predict['predict_date'].dt, f)
112 |         train_df[f] = getattr(train_df['record_date'].dt, f)
113 | 
114 | 
115 | # In[14]:
116 | 
117 | def extract_describe_feture(train_df, predict, f, date_span):
118 |     df = train_df.groupby(f).describe()['power_consumption'].unstack()
119 |     column_fmt = 'post{}_{}_{}'
120 |     df.columns = [column_fmt.format(date_span, f, x) for x in   df.columns]
121 |     predict = predict.join(df, on=f)
122 |     return predict
123 | 
124 | 
125 | # In[15]:
126 | 
127 | def extract_post_day_describe(date_span, train, predict, used_data_fetures, add_feture=True):
128 |     train_df = get_train_df(date_span, train)
129 |     if add_feture:
130 |         add_feture_in_date(used_data_fetures, predict, train_df)
131 |     for f in used_data_fetures:
132 |         predict = extract_describe_feture(train_df, predict, f, date_span)
133 |     return predict
134 |     
135 | 
136 | 
137 | # In[16]:
138 | 
139 | for date_span in time_spans:
140 |     predict = extract_post_day_describe(date_span, train, predict, used_data_fetures)
141 | 
142 | 
143 | # # 节假日特征 
144 | 
145 | # In[17]:
146 | 
147 | use_holiday_fetures = ['is_week', 'is_weekend', 'is_festival', 'is_holiday']
148 | 
149 | 
150 | # In[18]:
151 | 
152 | holiday_df = pd.read_csv('../dataset/holiday.csv', parse_dates=['date'])
153 | holiday_df.head(2)
154 | 
155 | 
156 | # In[19]:
157 | 
158 | holiday_df['is_holiday'] = 0
159 | 
160 | holiday_df.loc[holiday_df.holiday!=0, 'is_holiday'] = 1
161 | 
162 | 
163 | # In[20]:
164 | 
165 | df = pd.get_dummies(holiday_df.holiday)
166 | 
167 | 
168 | # In[21]:
169 | 
170 | df.columns = ['is_week', 'is_weekend', 'is_festival']
171 | 
172 | 
173 | # In[22]:
174 | 
175 | holiday_df = holiday_df.join(df)
176 | 
177 | 
178 | # In[23]:
179 | 
180 | holiday_df.drop('holiday', axis=1, inplace=True)
181 | 
182 | 
183 | # In[24]:
184 | 
185 | holiday_df.set_index('date', inplace=True)
186 | 
187 | 
188 | # #### 添加日期特征至训练集 
189 | 
190 | # In[25]:
191 | 
192 | train = train.join(holiday_df, on='record_date')
193 | predict = predict.join(holiday_df, on='predict_date')
194 | 
195 | 
196 | # In[26]:
197 | 
198 | for date_span in time_spans:
199 |     predict = extract_post_day_describe(
200 |         date_span,
201 |         train,
202 |         predict,
203 |         use_holiday_fetures,
204 |         add_feture=False)
205 | 
206 | 
207 | # ### 假期平均统计特征 
208 | 
209 | # In[27]:
210 | 
211 | def apply_describe_day_sum(group):
212 |     df = group.groupby('record_date')[['power_consumption']].sum().describe().unstack()['power_consumption']
213 |     return df
214 | 
215 | 
216 | # In[28]:
217 | 
218 | def extract_day_consumption_describe(span, f, train, predict, add_dt_feture=False):
219 |     tdf = get_train_df(span, train)
220 |     if add_dt_feture:
221 |         add_feture_in_date([f], predict, tdf)
222 |     name_fmt = 'post{}_{}_{}_day_consumption'
223 |     df = tdf.groupby(f).apply(apply_describe_day_sum)
224 |     df.columns = [name_fmt.format(span, f, x) for x in df.columns]
225 |     predict = predict.join(df, on=f)
226 |     return predict
227 | 
228 | 
229 | # In[29]:
230 | 
231 | # extract_day_consumption_describe(30, 'is_week', train, predict)
232 | 
233 | 
234 | # In[30]:
235 | 
236 | for f in use_holiday_fetures:
237 |     for span in time_spans:
238 |         if f != 'is_weekend':  # compelete same with is_week
239 |             predict = extract_day_consumption_describe(span, f, train, predict)
240 | 
241 | 
242 | # # 总数统计特征 
243 | 
244 | # In[31]:
245 | 
246 | used_sum_fetures = [
247 |     'dayofweek', 'dayofyear', 'days_in_month', 
248 |     'quarter', 'week', 
249 |     'month', 'year'
250 | ]
251 | 
252 | 
253 | # In[32]:
254 | 
255 | def extract_mean_consumption(train_df, feture, date_span, predict, div_times):
256 |     column_fmt = 'post{}_{}_mean_consumption'
257 |     df = train_df.groupby(feture)['power_consumption'].sum() / (date_span / div_times)
258 |     df.name = column_fmt.format(date_span, feture)
259 |     predict = predict.join(df)
260 |     return predict
261 | 
262 | 
263 | # In[33]:
264 | 
265 | def extract_post_mean_consumption(feture, div_times, date_span, train, predict):
266 |     train_df = get_train_df(30, train)
267 |     add_feture_in_date(used_sum_fetures, predict, train_df)
268 |     predict = extract_mean_consumption(train_df, feture, date_span, predict, div_times)
269 |     return predict
270 | 
271 | 
272 | # In[34]:
273 | 
274 | def extract_all_post_mean_consumption(used_sum_feture_model, train, predict):   
275 |     for feture, div_times, date_spans in used_sum_feture_model:
276 |         for date_span in date_spans:
277 |             predict = extract_post_mean_consumption(feture, div_times, date_span, train, predict)
278 |     return predict
279 | 
280 | 
281 | # In[35]:
282 | 
283 | used_sum_feture_model = [
284 |     ('dayofweek', 7, [30, 60, 90, 180, 360]),
285 |     ('dayofyear', 1, [30, 60, 90, 180, 360]),
286 |     ('days_in_month', 30, [30, 60, 90, 180, 360]),
287 |     ('quarter', 90, [90, 180, 360]),
288 |     ('week', 52, [180, 360]),
289 |     ('month', 30, [30, 60, 90, 120, 240, 360]),
290 |     ('year', 360, [ 360]), 
291 | ]
292 | 
293 | 
294 | # In[36]:
295 | 
296 | # predict = extract_all_post_mean_consumption(used_sum_feture_model, train, predict)  # over fit
297 | 
298 | 
299 | # In[37]:
300 | 
301 | # extract_day_consumption_describe(30, 'dayofweek', train, predict, add_dt_feture=True)
302 | 
303 | 
304 | # In[38]:
305 | 
306 | for f in used_sum_fetures:
307 |     for span in time_spans:
308 |         predict = extract_day_consumption_describe(span, f, train, predict, add_dt_feture=True)
309 | 
310 | 
311 | # # 天气特征 
312 | 
313 | # In[39]:
314 | 
315 | header = ['weather_date', 'weather_max', 'weather_min', 'weather_type', 'weather_wind', 'wind_type']
316 | 
317 | 
318 | # In[40]:
319 | 
320 | weather_df = pd.read_csv('../dataset/yangzhong.csv', header=None, names=header, parse_dates=['weather_date'])
321 | weather_df.head(3)
322 | 
323 | 
324 | # #### 将天气切分成块,再提取块边界 
325 | 
326 | # In[41]:
327 | 
328 | weather_df.weather_min = pd.cut(weather_df.weather_min, bins=10)
329 | 
330 | 
331 | # In[42]:
332 | 
333 | weather_df.weather_min = weather_df.weather_min.str.extract('\((-?\d+\.?\d*),').astype(np.float)
334 | 
335 | 
336 | # In[43]:
337 | 
338 | weather_df.weather_max = pd.cut(weather_df.weather_max, bins=10)
339 | 
340 | 
341 | # In[44]:
342 | 
343 | weather_df.weather_max = weather_df.weather_max.str.extract(', (-?\d+\.?\d*)\]').astype(np.float)
344 | 
345 | 
346 | # In[45]:
347 | 
348 | weather_df.shape
349 | 
350 | 
351 | # In[46]:
352 | 
353 | weather_type_count = weather_df.weather_type.value_counts()
354 | 
355 | 
356 | # In[47]:
357 | 
358 | weather_df.weather_type = weather_df.weather_type.replace([
359 |     x for x in weather_type_count.loc[weather_type_count < 2].index
360 | ], 'rare_weather')
361 | 
362 | 
363 | # In[48]:
364 | 
365 | weather_df.loc[weather_df.weather_type.str.contains('阵雨'), 'weather_type'] = 'showers_weather'
366 | 
367 | 
368 | # In[49]:
369 | 
370 | weather_df.loc[weather_df.weather_type.str.contains('雨'), 'weather_type'] = 'rain_weather'
371 | 
372 | 
373 | # In[50]:
374 | 
375 | weather_df.loc[~weather_df.weather_type.str.islower(), 'weather_type'] = 'fine_weather'
376 | 
377 | 
378 | # In[51]:
379 | 
380 | weather_df = weather_df.join(pd.get_dummies(weather_df.weather_type))
381 | 
382 | 
383 | # In[75]:
384 | 
385 | weather_num_columns = [
386 |     'weather_max', 'weather_min', 'wind_type', 'fine_weather', 'rain_weather',
387 |     'rare_weather', 'showers_weather'
388 | ]
389 | 
390 | 
391 | # In[53]:
392 | 
393 | weather_df.head()
394 | 
395 | 
396 | # In[54]:
397 | 
398 | weather_df.weather_wind.replace('东南风', 'southeast_wind', inplace=True)
399 | weather_df.weather_wind.replace('东北风', 'northeast_wind', inplace=True)
400 | weather_df.weather_wind.replace('西南风', 'southwest_wind', inplace=True)
401 | weather_df.weather_wind.replace('西北风', 'northwest_wind', inplace=True)
402 | weather_df.weather_wind.replace('东风', 'east_wind', inplace=True)
403 | weather_df.weather_wind.replace('北风', 'north_wind', inplace=True)
404 | weather_df.weather_wind.replace('南风', 'south_wind', inplace=True)
405 | weather_df.weather_wind.replace('西风', 'west_wind', inplace=True)
406 | 
407 | 
408 | # In[55]:
409 | 
410 | weather_df.weather_wind.replace(['3-4级', '暂无实况', '无持续风向'], 'unknow_wind', inplace=True)
411 | 
412 | 
413 | # In[56]:
414 | 
415 | weather_df = weather_df.join(pd.get_dummies(weather_df.weather_wind))
416 | 
417 | 
418 | # In[57]:
419 | 
420 | wind_type_count = weather_df.wind_type.value_counts()
421 | 
422 | 
423 | # In[58]:
424 | 
425 | weather_df.wind_type = weather_df.wind_type.replace([
426 |     x for x in wind_type_count.loc[wind_type_count < 5 ].index
427 | ], 'rare_wind')
428 | 
429 | 
430 | # In[59]:
431 | 
432 | weather_df.wind_type.replace('rare_wind', 0, inplace=True)
433 | weather_df.wind_type.replace('微风', 1, inplace=True)
434 | weather_df.wind_type.replace('1级', 2, inplace=True)
435 | weather_df.wind_type.replace('2级', 3, inplace=True)
436 | weather_df.wind_type.replace('小于3级', 4, inplace=True)
437 | weather_df.wind_type.replace('3级', 6, inplace=True)
438 | weather_df.wind_type.replace('3-4级转小于3级', 5, inplace=True)
439 | weather_df.wind_type.replace('3-4级', 7, inplace=True)
440 | weather_df.wind_type.replace('4-5级转3-4级', 8, inplace=True)
441 | weather_df.wind_type.replace('4-5级', 9, inplace=True)
442 | 
443 | 
444 | # In[60]:
445 | 
446 | weather_df.wind_type.value_counts()
447 | 
448 | 
449 | # In[61]:
450 | 
451 | weather_df.head()
452 | 
453 | 
454 | # In[62]:
455 | 
456 | train = train.join(weather_df.set_index('weather_date'), on='record_date')
457 | 
458 | 
459 | # In[63]:
460 | 
461 | predict = predict.join(weather_df.set_index('weather_date'), on='predict_date')
462 | 
463 | 
464 | # ### 添加天气特征(超前) 
465 | 
466 | # In[64]:
467 | 
468 | use_weather_fetures = ['weather_max', 'weather_min', 'weather_type', 'weather_wind', 'wind_type']
469 | 
470 | 
471 | # In[65]:
472 | 
473 | for date_span in time_spans:
474 |     predict = extract_post_day_describe(
475 |         date_span,
476 |         train,
477 |         predict,
478 |         use_weather_fetures,
479 |         add_feture=False)
480 | 
481 | 
482 | # In[66]:
483 | 
484 | # extract_day_consumption_describe(30, 'weather_max', train, predict, add_dt_feture=False)
485 | 
486 | 
487 | # In[67]:
488 | 
489 | for f in use_weather_fetures:
490 |     for span in time_spans:
491 |         predict = extract_day_consumption_describe(span, f, train, predict, add_dt_feture=False)
492 | 
493 | 
494 | # # 交叉特征
495 | 
496 | # ##  时间 X 假期
497 | 
498 | # In[73]:
499 | 
500 | def extract_combin_feture_day_consumption_describe(span, f1, f2, predict, train, f1_need_add=True, f2_need_add=False):
501 |     tdf = get_train_df(span, train)
502 |     if f1_need_add:
503 |         add_feture_in_date([f1], predict, tdf)
504 |     if f2_need_add:
505 |         add_feture_in_date([f2], predict, tdf) 
506 |     new_feture_name = 'post{}_combine_{}_and_{}'.format(span, f1, f2)
507 |     predict[new_feture_name] = predict[f1] * predict[f2]
508 |     tdf[new_feture_name] = tdf[f1] * tdf[f2]
509 |     predict = extract_day_consumption_describe(span, new_feture_name, tdf, predict)
510 |     return predict
511 | 
512 | 
513 | # In[69]:
514 | 
515 | # extract_combin_feture_day_consumption_describe(30, 'dayofweek', 'is_week', predict, train)
516 | 
517 | 
518 | # In[74]:
519 | 
520 | for span in time_spans:
521 |     for f1 in used_data_fetures:
522 |         for f2 in use_holiday_fetures:
523 |             predict = extract_combin_feture_day_consumption_describe(span, f1, f2, predict, train)
524 | 
525 | 
526 | # ## 时间 X 天气 
527 | 
528 | # In[78]:
529 | 
530 | for span in time_spans:
531 |     for f1 in used_data_fetures:
532 |         for f2 in weather_num_columns:
533 |             predict = extract_combin_feture_day_consumption_describe(span, f1, f2, predict, train)
534 | 
535 | 
536 | # ## 天气 X 假期
537 | 
538 | # In[79]:
539 | 
540 | for span in time_spans:
541 |     for f1 in weather_num_columns:
542 |         for f2 in use_holiday_fetures:
543 |             predict = extract_combin_feture_day_consumption_describe(span, f1, f2, predict, train, False, False)
544 | 
545 | 
546 | # In[ ]:
547 | 
548 | not_use_columns = ['weather_type', 'weather_wind']
549 | 
550 | 
551 | # In[ ]:
552 | 
553 | # data_feture_not_use = [x for x in used_data_fetures if  x.startswith('is_')]
554 | 
555 | 
556 | # In[ ]:
557 | 
558 | feture_columns = [x for x in predict.columns if x not in not_use_columns ]
559 | 
560 | 
561 | # In[ ]:
562 | 
563 | # feture_columns = [x for x in feture_columns if x not in use_weather_fetures]
564 | 
565 | 
566 | # In[ ]:
567 | 
568 | # feture_columns = [x for x in feture_columns if x not in use_holiday_fetures]
569 | 
570 | 
571 | # In[ ]:
572 | 
573 | 'weather_max' in feture_columns, 'is_week' in feture_columns
574 | 
575 | 
576 | # In[ ]:
577 | 
578 | predict.to_csv(feture_path, columns=feture_columns, index=False)
579 | 
580 | 


--------------------------------------------------------------------------------
/code/holiday_crawl.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 54,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import requests\n",
 12 |     "import pandas as pd\n",
 13 |     "import json\n",
 14 |     "from os.path import exists"
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "code",
 19 |    "execution_count": 59,
 20 |    "metadata": {
 21 |     "collapsed": true
 22 |    },
 23 |    "outputs": [],
 24 |    "source": [
 25 |     "def write_to_file(filename, data):\n",
 26 |     "    if not exists(filename):\n",
 27 |     "        with open(filename, 'w') as f:\n",
 28 |     "            f.write('date,holiday\\n')   \n",
 29 |     "    for k, v in json.loads(data).items():\n",
 30 |     "        if type(v) == str:\n",
 31 |     "            v = int(v)\n",
 32 |     "        with open(filename, 'a') as f:\n",
 33 |     "            f.write('{},{}\\n'.format(k, v))        "
 34 |    ]
 35 |   },
 36 |   {
 37 |    "cell_type": "code",
 38 |    "execution_count": 2,
 39 |    "metadata": {
 40 |     "collapsed": false
 41 |    },
 42 |    "outputs": [],
 43 |    "source": [
 44 |     "data = pd.date_range('2015/1/1', '2016/12/31')"
 45 |    ]
 46 |   },
 47 |   {
 48 |    "cell_type": "code",
 49 |    "execution_count": 3,
 50 |    "metadata": {
 51 |     "collapsed": false
 52 |    },
 53 |    "outputs": [],
 54 |    "source": [
 55 |     "date_ags = data.map(lambda x:x.strftime('%Y%m%d'))"
 56 |    ]
 57 |   },
 58 |   {
 59 |    "cell_type": "code",
 60 |    "execution_count": 19,
 61 |    "metadata": {
 62 |     "collapsed": true
 63 |    },
 64 |    "outputs": [],
 65 |    "source": [
 66 |     "url_fmt = 'http://tool.bitefu.net/jiari/?d={}'"
 67 |    ]
 68 |   },
 69 |   {
 70 |    "cell_type": "code",
 71 |    "execution_count": 30,
 72 |    "metadata": {
 73 |     "collapsed": false
 74 |    },
 75 |    "outputs": [],
 76 |    "source": [
 77 |     "agrv = ','.join(date_ags[:100])"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": 31,
 83 |    "metadata": {
 84 |     "collapsed": false
 85 |    },
 86 |    "outputs": [
 87 |     {
 88 |      "data": {
 89 |       "text/plain": [
 90 |        "'http://tool.bitefu.net/jiari/?d=20150101,20150102,20150103,20150104,20150105,20150106,20150107,20150108,20150109,20150110,20150111,20150112,20150113,20150114,20150115,20150116,20150117,20150118,20150119,20150120,20150121,20150122,20150123,20150124,20150125,20150126,20150127,20150128,20150129,20150130,20150131,20150201,20150202,20150203,20150204,20150205,20150206,20150207,20150208,20150209,20150210,20150211,20150212,20150213,20150214,20150215,20150216,20150217,20150218,20150219,20150220,20150221,20150222,20150223,20150224,20150225,20150226,20150227,20150228,20150301,20150302,20150303,20150304,20150305,20150306,20150307,20150308,20150309,20150310,20150311,20150312,20150313,20150314,20150315,20150316,20150317,20150318,20150319,20150320,20150321,20150322,20150323,20150324,20150325,20150326,20150327,20150328,20150329,20150330,20150331,20150401,20150402,20150403,20150404,20150405,20150406,20150407,20150408,20150409,20150410'"
 91 |       ]
 92 |      },
 93 |      "execution_count": 31,
 94 |      "metadata": {},
 95 |      "output_type": "execute_result"
 96 |     }
 97 |    ],
 98 |    "source": [
 99 |     "url = url_fmt.format(agrv)\n",
100 |     "url"
101 |    ]
102 |   },
103 |   {
104 |    "cell_type": "code",
105 |    "execution_count": 63,
106 |    "metadata": {
107 |     "collapsed": false
108 |    },
109 |    "outputs": [],
110 |    "source": [
111 |     "for i in range(0, date_ags.shape[0], 100):\n",
112 |     "    agrv = ','.join(date_ags[i:(i+100)])\n",
113 |     "    url = url_fmt.format(agrv)\n",
114 |     "    res = requests.get(url)\n",
115 |     "    write_to_file('../dataset/holiday.csv', res.text)"
116 |    ]
117 |   }
118 |  ],
119 |  "metadata": {
120 |   "anaconda-cloud": {},
121 |   "hide_input": false,
122 |   "kernelspec": {
123 |    "display_name": "Python 3",
124 |    "language": "python",
125 |    "name": "python3"
126 |   },
127 |   "language_info": {
128 |    "codemirror_mode": {
129 |     "name": "ipython",
130 |     "version": 3
131 |    },
132 |    "file_extension": ".py",
133 |    "mimetype": "text/x-python",
134 |    "name": "python",
135 |    "nbconvert_exporter": "python",
136 |    "pygments_lexer": "ipython3",
137 |    "version": "3.5.1"
138 |   }
139 |  },
140 |  "nbformat": 4,
141 |  "nbformat_minor": 2
142 | }
143 | 


--------------------------------------------------------------------------------
/code/predict_result.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [
 10 |     {
 11 |      "name": "stderr",
 12 |      "output_type": "stream",
 13 |      "text": [
 14 |       "/home/zhanglun/.pyenv/versions/anaconda3-4.1.0/lib/python3.5/site-packages/sklearn/cross_validation.py:44: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
 15 |       "  \"This module will be removed in 0.20.\", DeprecationWarning)\n"
 16 |      ]
 17 |     }
 18 |    ],
 19 |    "source": [
 20 |     "import pickle\n",
 21 |     "\n",
 22 |     "import numpy as np\n",
 23 |     "import scipy as sp\n",
 24 |     "import pandas as pd\n",
 25 |     "\n",
 26 |     "from sklearn.ensemble import RandomForestRegressor\n",
 27 |     "from sklearn.metrics import r2_score, roc_auc_score\n",
 28 |     "from sklearn import cross_validation\n",
 29 |     "from sklearn.model_selection import GridSearchCV\n",
 30 |     "from sklearn.svm import SVR "
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": 2,
 36 |    "metadata": {
 37 |     "collapsed": true
 38 |    },
 39 |    "outputs": [],
 40 |    "source": [
 41 |     "model_path = '../dataset/fetures/model.pkl'\n",
 42 |     "output_path = '../submit/Tianchi_power_predict_table.csv'"
 43 |    ]
 44 |   },
 45 |   {
 46 |    "cell_type": "code",
 47 |    "execution_count": 3,
 48 |    "metadata": {
 49 |     "collapsed": true
 50 |    },
 51 |    "outputs": [],
 52 |    "source": [
 53 |     "fetures = pd.read_csv('../dataset/fetures/test_feture.csv', parse_dates=['predict_date'])"
 54 |    ]
 55 |   },
 56 |   {
 57 |    "cell_type": "code",
 58 |    "execution_count": 4,
 59 |    "metadata": {
 60 |     "collapsed": true
 61 |    },
 62 |    "outputs": [],
 63 |    "source": [
 64 |     "model = pickle.load(open(model_path, 'rb'))"
 65 |    ]
 66 |   },
 67 |   {
 68 |    "cell_type": "code",
 69 |    "execution_count": 5,
 70 |    "metadata": {
 71 |     "collapsed": false
 72 |    },
 73 |    "outputs": [],
 74 |    "source": [
 75 |     "train_x = fetures.drop(['predict_date', 'predict_power_consumption'], axis=1)"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 6,
 81 |    "metadata": {
 82 |     "collapsed": false
 83 |    },
 84 |    "outputs": [],
 85 |    "source": [
 86 |     "fetures.predict_power_consumption = model.predict(train_x.fillna(0)).astype(np.int64)"
 87 |    ]
 88 |   },
 89 |   {
 90 |    "cell_type": "code",
 91 |    "execution_count": 7,
 92 |    "metadata": {
 93 |     "collapsed": false
 94 |    },
 95 |    "outputs": [],
 96 |    "source": [
 97 |     "fetures.predict_date = fetures.predict_date.map(lambda x: x.strftime('%Y%m%d'))"
 98 |    ]
 99 |   },
100 |   {
101 |    "cell_type": "code",
102 |    "execution_count": 8,
103 |    "metadata": {
104 |     "collapsed": false
105 |    },
106 |    "outputs": [
107 |     {
108 |      "data": {
109 |       "text/plain": [
110 |        "3754006.806451613"
111 |       ]
112 |      },
113 |      "execution_count": 8,
114 |      "metadata": {},
115 |      "output_type": "execute_result"
116 |     }
117 |    ],
118 |    "source": [
119 |     "fetures['predict_power_consumption'].mean()"
120 |    ]
121 |   },
122 |   {
123 |    "cell_type": "code",
124 |    "execution_count": 9,
125 |    "metadata": {
126 |     "collapsed": true
127 |    },
128 |    "outputs": [],
129 |    "source": [
130 |     "fetures.to_csv(output_path, index=False, columns=['predict_date', 'predict_power_consumption'])"
131 |    ]
132 |   }
133 |  ],
134 |  "metadata": {
135 |   "anaconda-cloud": {},
136 |   "hide_input": false,
137 |   "kernelspec": {
138 |    "display_name": "Python 3",
139 |    "language": "python",
140 |    "name": "python3"
141 |   },
142 |   "language_info": {
143 |    "codemirror_mode": {
144 |     "name": "ipython",
145 |     "version": 3
146 |    },
147 |    "file_extension": ".py",
148 |    "mimetype": "text/x-python",
149 |    "name": "python",
150 |    "nbconvert_exporter": "python",
151 |    "pygments_lexer": "ipython3",
152 |    "version": "3.5.1"
153 |   }
154 |  },
155 |  "nbformat": 4,
156 |  "nbformat_minor": 2
157 | }
158 | 


--------------------------------------------------------------------------------
/code/result.txt:
--------------------------------------------------------------------------------
 1 | 2017-06-10 17:56:11.519155 result: 0.9809955701469716 fetures: 702 post360_weekofyear_std:0.05983807806632941  , post360_week_std:0.05727341629026522  , post240_dayofyear_std:0.05700243617276306  , post360_dayofyear_75%:0.05470857403513062  , post360_dayofyear_std:0.05276191776918991  , post240_dayofyear_75%:0.05024778442182545  , post120_dayofyear_50%:0.04432864932493469  , post120_dayofyear_25%:0.030696122647652965  , post240_dayofyear_max:0.029791901464045943  , post360_dayofyear_max:0.029125100472333708  , post360_weekofyear_count:0.02428621103910183  , post360_week_count:0.024117039219537808  , post240_dayofyear_50%:0.023653997600368914  , post360_weekofyear_25%:0.02358902686612497  , post360_weekofyear_max:0.02339527645346069  , post360_week_25%:0.02167830480513648  , post360_dayofyear_50%:0.020969892112941615  , post360_weekofyear_mean:0.020865596411447317  , post360_week_mean:0.020820565163031164  , post240_week_std:0.020408987363339584 
 2 | 2017-06-10 17:56:57.531889 result: 0.9809955701469716 fetures: 702 post360_weekofyear_std:0.05983807806632941  , post360_week_std:0.05727341629026522  , post240_dayofyear_std:0.05700243617276306  , post360_dayofyear_75%:0.05470857403513062  , post360_dayofyear_std:0.05276191776918991  , post240_dayofyear_75%:0.05024778442182545  , post120_dayofyear_50%:0.04432864932493469  , post120_dayofyear_25%:0.030696122647652965  , post240_dayofyear_max:0.029791901464045943  , post360_dayofyear_max:0.029125100472333708  , post360_weekofyear_count:0.02428621103910183  , post360_week_count:0.024117039219537808  , post240_dayofyear_50%:0.023653997600368914  , post360_weekofyear_25%:0.02358902686612497  , post360_weekofyear_max:0.02339527645346069  , post360_week_25%:0.02167830480513648  , post360_dayofyear_50%:0.020969892112941615  , post360_weekofyear_mean:0.020865596411447317  , post360_week_mean:0.020820565163031164  , post240_week_std:0.020408987363339584 
 3 | 2017-06-10 17:58:23.692043 result: 0.9812260591191833 fetures: 702 post240_dayofyear_std:0.06061743530771681  , post360_dayofyear_75%:0.05758784744580659  , post360_weekofyear_std:0.05700257710150259  , post360_week_std:0.05475886198817952  , post120_dayofyear_50%:0.04903319222276939  , post240_dayofyear_75%:0.04740758625355467  , post360_dayofyear_std:0.04045348492237613  , post240_dayofyear_max:0.03420164113736823  , post120_dayofyear_25%:0.031313071544771066  , post360_week_mean:0.029646691307865077  , post240_dayofyear_50%:0.027636178147651215  , post360_dayofyear_max:0.027471913720788735  , post360_week_count:0.02657281098445252  , post360_week_max:0.025361665971013903  , post360_weekofyear_25%:0.025076057967035353  , post360_weekofyear_max:0.02439508472114531  , post360_weekofyear_mean:0.024281116428763357  , post360_weekofyear_count:0.022381714522338794  , post360_dayofyear_50%:0.01977948438253097  , post120_dayofyear_std:0.019053914376049304 
 4 | 2017-06-10 18:07:38.840265 result: (0.9812260591191833, 0.95092811109387121, 0.61914474576265377, 0.71653278548142452, 0.43749602639182639) fetures: 702 post240_dayofyear_std:0.06061743530771681  , post360_dayofyear_75%:0.05758784744580659  , post360_weekofyear_std:0.05700257710150259  , post360_week_std:0.05475886198817952  , post120_dayofyear_50%:0.04903319222276939  , post240_dayofyear_75%:0.04740758625355467  , post360_dayofyear_std:0.04045348492237613  , post240_dayofyear_max:0.03420164113736823  , post120_dayofyear_25%:0.031313071544771066  , post360_week_mean:0.029646691307865077  , post240_dayofyear_50%:0.027636178147651215  , post360_dayofyear_max:0.027471913720788735  , post360_week_count:0.02657281098445252  , post360_week_max:0.025361665971013903  , post360_weekofyear_25%:0.025076057967035353  , post360_weekofyear_max:0.02439508472114531  , post360_weekofyear_mean:0.024281116428763357  , post360_weekofyear_count:0.022381714522338794  , post360_dayofyear_50%:0.01977948438253097  , post120_dayofyear_std:0.019053914376049304 
 5 | 2017-06-10 18:13:41.316884 result: (0.9815575664129459, 0.94995628600124637, 0.61416807863600109, 0.75009225632557885, 0.40984849907331433) fetures: 702 post240_dayofyear_std:0.07093754904341293  , post360_dayofyear_75%:0.061197642670642656  , post360_weekofyear_std:0.05263520169823813  , post360_dayofyear_std:0.04887795505499953  , post360_week_std:0.04622601588097645  , post240_dayofyear_75%:0.04489649678455263  , post120_dayofyear_50%:0.040187236168927845  , post240_dayofyear_max:0.0369892481419167  , post360_weekofyear_max:0.034461899665024394  , post360_week_mean:0.0340585407347569  , post360_week_max:0.031065407256356273  , post360_week_count:0.030808708306986312  , post240_dayofyear_50%:0.025359374251381593  , post360_weekofyear_mean:0.025299285022859355  , post120_dayofyear_std:0.0219081681188889  , post360_weekofyear_count:0.021401989788210073  , post360_dayofyear_max:0.021353503033307436  , post360_weekofyear_25%:0.02073365144202253  , post240_weekofyear_std:0.020610251119400843  , post120_dayofyear_25%:0.019286342292937124 
 6 | 2017-06-10 21:15:10.157383 result: (0.9822051009352075, 0.95579364610049611, 0.7062930556330822, 0.81414454969936811, 0.56954695479573703) fetures: 898 post360_week_std:0.06762393904927758  , post240_dayofyear_std:0.05325587110851327  , post360_dayofyear_75%:0.05136696042942002  , post360_dayofyear_std:0.04943380259543345  , post360_weekofyear_std:0.04934004811353454  , post240_dayofyear_75%:0.04597445495162306  , post120_dayofyear_50%:0.044688318382925515  , post360_week_25%:0.02948095048177686  , post240_dayofyear_max:0.0289554116986942  , post360_week_count:0.02653548392148828  , post120_dayofyear_25%:0.02497912510445329  , is_festival:0.02482572307566017  , post360_weekofyear_25%:0.02323131702891193  , post360_week_mean:0.0232217056392544  , post360_dayofyear_max:0.022677801111633936  , is_week:0.020851448143867687  , post360_week_max:0.020792355477846667  , is_holiday:0.020423909005792922  , post360_weekofyear_max:0.020297191740446426  , post360_dayofyear_50%:0.020162771299375385 
 7 | 2017-06-10 21:36:00.079713 result: (0.9826812518986928, 0.95605369543960583, 0.68937185194605832, 0.8199889398361534, 0.52346412256255381) fetures: 868 post360_week_std:0.059184042232527816  , post360_dayofyear_std:0.05783583438730425  , post240_dayofyear_std:0.057574260680706646  , post240_dayofyear_75%:0.05160454714145169  , post360_week_max:0.045137841843380204  , post360_dayofyear_75%:0.04227164346666081  , post120_dayofyear_50%:0.03569903264919281  , post360_week_count:0.035174477706062024  , post360_weekofyear_std:0.03222649606930148  , post360_dayofyear_max:0.031951271707399885  , post360_weekofyear_count:0.031013369532032663  , post360_weekofyear_50%:0.030265111090275347  , post360_week_25%:0.028118094560222056  , post360_weekofyear_mean:0.026757913582540902  , is_festival:0.024883471277168966  , post240_dayofyear_max:0.02439109468279022  , post360_weekofyear_max:0.021607394475592873  , post240_dayofyear_50%:0.021476271530959577  , is_week:0.020138534061675803  , is_holiday:0.020072241653007562 
 8 | 2017-06-10 22:44:09.230867 result: (0.9810693443654728, 0.94978575722172542, 0.62406609817483327, 0.72017865084301147, 0.46713572931636471) fetures: 1104 post360_weekofyear_std:0.06948627102325142  , post240_dayofyear_std:0.05567522583766983  , post360_dayofyear_std:0.05323543240992519  , post240_dayofyear_75%:0.0517225978339905  , post360_dayofyear_75%:0.05171704142873725  , post120_dayofyear_50%:0.05020134036423885  , post360_week_std:0.049976129308657434  , post240_dayofyear_max:0.03095550481825238  , post360_week_25%:0.030547290124186784  , post360_dayofyear_max:0.029104558549925125  , post360_week_count:0.02797305019800481  , post120_dayofyear_25%:0.027965956286967103  , post360_week_mean:0.02626641749445518  , post240_dayofyear_50%:0.02226705197443881  , post360_weekofyear_mean:0.022078075749952266  , post360_dayofyear_50%:0.021391665624325422  , post360_weekofyear_25%:0.0205594437435275  , post240_weekofyear_std:0.01971365781031094  , post240_week_std:0.018831011385202025  , post360_week_max:0.018569535554401162 
 9 | 2017-06-10 22:54:10.250829 result: (0.982363706195949, 0.96066600069520058, 0.64939188461289943, 0.80822338640587588, 0.22162859500331256) fetures: 2208 post150_dayofyear_max:0.16010911457429855  , post150_dayofyear_std:0.15292631108579458  , post360_weather_max_count:0.037645671153132575  , post240_weather_max_std:0.03237362057319401  , post150_weather_max_max:0.027552593112726082  , post270_weather_max_std:0.025471248648629628  , post300_weather_max_std:0.02440157378458934  , post120_dayofyear_50%:0.020028400502903384  , post210_weather_max_std:0.017888068949600374  , post150_dayofyear_mean:0.017564641077130197  , post330_weather_max_std:0.013859076623724549  , post150_weather_max_std:0.011204284738801036  , post330_weather_max_count:0.01059584319752082  , post150_weather_max_count:0.009314193188305989  , post360_weekofyear_25%:0.008092697472318493  , post60_weather_max_std:0.00780814359692211  , post120_dayofyear_75%:0.007589314103314759  , post180_weather_max_std:0.0075460925315592575  , post240_weekofyear_std:0.007102608588847814  , post150_weather_max_mean:0.006962719280855951 
10 | 2017-06-10 23:06:36.005611 result: (0.9824143773145466, 0.95931549164187524, 0.64046212967688765, 0.79641664067145457, 0.38358848051457639) fetures: 1728 post150_dayofyear_max:0.24589109189856068  , post150_dayofyear_std:0.22125646763239545  , post150_dayofyear_mean:0.03152945721600716  , post120_dayofyear_50%:0.016989123579456967  , post270_dayofyear_75%:0.01459198182447194  , post360_weekofyear_std:0.013079798772271058  , post360_week_max:0.011882143256492665  , post360_weekofyear_max:0.011258373307040648  , post240_weekofyear_std:0.011240134153628425  , post360_dayofyear_75%:0.009749179635727679  , post360_week_25%:0.009643561204603125  , post300_dayofyear_75%:0.008596215464555645  , post210_dayofyear_75%:0.008455530194865784  , post360_weekofyear_mean:0.008303488957485871  , post180_dayofyear_50%:0.00819432444982622  , post300_week_max:0.007911045044612583  , post150_dayofyear_50%:0.007773686877818734  , post150_dayofyear_25%:0.007649709348662322  , post180_dayofyear_75%:0.0075052377317228365  , post360_dayofyear_max:0.007393453393643394 
11 | 2017-06-10 23:15:27.964041 result: (0.9817043822495296, 0.95580068018658104, 0.66005799770865203, 0.81533845753226031, 0.40983835558976345) fetures: 2208 post150_dayofyear_max:0.25638705349450625  , post150_dayofyear_std:0.22165863246951925  , post150_dayofyear_mean:0.01789594183099462  , post360_week_std:0.01494142111616415  , post120_dayofyear_50%:0.013747571545445494  , post150_dayofyear_25%:0.012606159293074258  , post360_weekofyear_std:0.012412248779932702  , post210_dayofyear_75%:0.011140208785617308  , post180_dayofyear_75%:0.011015741245508671  , post240_dayofyear_75%:0.00986211952643449  , post360_weekofyear_max:0.009666276834181161  , post300_dayofyear_75%:0.009558421526013045  , post360_dayofyear_75%:0.009319439013739243  , post270_dayofyear_75%:0.007669082305535447  , post330_dayofyear_75%:0.007541019805731144  , post360_week_25%:0.007126927259613465  , post180_dayofyear_25%:0.006882667588199024  , post300_dayofyear_50%:0.00659357818679421  , post360_week_max:0.006537911584139208  , post300_week_max:0.006438391179657396 
12 | 2017-06-10 23:21:32.236102 result: (0.9824506245857773, 0.95924409064056793, 0.65242749790927546, 0.80825911122580352, 0.39922845296404402) fetures: 2208 post150_dayofyear_max:0.24005512463558326  , post150_dayofyear_std:0.2308148293765062  , post150_dayofyear_mean:0.03229727832730095  , post120_dayofyear_50%:0.016920113130830892  , post360_weekofyear_std:0.014667239930494982  , post180_dayofyear_75%:0.013149619830856779  , post360_week_max:0.012963540123670111  , post300_dayofyear_75%:0.011529695803939392  , post210_dayofyear_75%:0.009928935148667488  , post360_week_25%:0.009561354169231642  , post360_weekofyear_max:0.009214851291084073  , post360_dayofyear_75%:0.008817977596810429  , post210_dayofyear_50%:0.008698094521734645  , post360_weekofyear_mean:0.007794007459013409  , post240_dayofyear_75%:0.0076514494492410075  , post300_weekofyear_max:0.007548760921191867  , post240_weekofyear_std:0.00735092696081901  , post300_week_max:0.007214018895017005  , post270_dayofyear_75%:0.007138834537515316  , post180_dayofyear_max:0.00706807903131027 
13 | 2017-06-10 23:45:42.921774 result: (0.9831875193825498, 0.95945394893937186, 0.686092964194567, 0.81812145631897049, 0.4700743846651656) (0.9831875193825498, 0.95945394893937186, 0.686092964194567, 0.81812145631897049, 0.4700743846651656) params: {'criterion': 'mae', 'max_features': 100, 'n_estimators': 100} 0.702487671874 fetures: 2208 post150_dayofyear_std:0.0330885869833952  , post150_dayofyear_max:0.029283462573992693  , post150_dayofyear_25%:0.021590656211323757  , post300_dayofyear_std:0.01771296856613513  , post360_weekofyear_max:0.017562392536192028  , post360_week_std:0.01662561736452066  , post360_weekofyear_mean:0.016408935057646807  , post180_dayofyear_std:0.015501636268463981  , post360_dayofyear_std:0.015005344722218634  , post150_dayofyear_mean:0.014911514712743516  , post120_dayofyear_75%:0.013539354778823016  , post210_dayofyear_std:0.013456791749897508  , post360_dayofyear_max:0.01321322947382893  , post210_dayofyear_75%:0.012718799593584563  , post360_week_count:0.011449106393858403  , post180_dayofyear_50%:0.011430833343111053  , post360_weekofyear_std:0.01111861692907194  , post300_dayofyear_75%:0.010965035796560068  , post330_dayofyear_std:0.01087378143369563  , post240_dayofyear_50%:0.010644420580679303 
14 | 2017-06-10 23:53:26.929485 result: (0.9824345308664283, 0.95767507375570637, 0.68518882772904433, 0.81661477804353044, 0.44971268884566817) (0.9824345308664283, 0.95767507375570637, 0.68518882772904433, 0.81661477804353044, 0.44971268884566817) params: {'criterion': 'mse', 'n_estimators': 300, 'max_features': 'sqrt'} 0.70941280102 fetures: 2208 post150_dayofyear_max:0.0256325632387379  , post150_dayofyear_std:0.022164739381156744  , post150_dayofyear_25%:0.01887170334889061  , post360_week_count:0.013925713673282339  , post360_weekofyear_25%:0.013687948386007011  , post210_dayofyear_std:0.013512463760180381  , post360_weekofyear_std:0.013426669767689504  , post360_weekofyear_max:0.012316258668840058  , post120_dayofyear_75%:0.01225970144645653  , post330_dayofyear_max:0.012190256572543074  , post150_dayofyear_50%:0.011958771056427002  , post360_dayofyear_std:0.011845140641975187  , post150_dayofyear_mean:0.01171720352493833  , post300_dayofyear_std:0.011477847650876224  , post360_dayofyear_max:0.01133734649391173  , post150_dayofyear_75%:0.011327980674041449  , post360_week_std:0.011303593156636498  , post360_weekofyear_50%:0.011248646457708645  , post360_week_75%:0.011075548365801196  , post180_dayofyear_std:0.010474519204089896 
15 | 2017-06-11 06:52:37.243021 result: (0.9831683882309213, 0.95980503571601361, 0.68308778066567644, 0.83110382797538129, 0.47037101851499641) params: {'criterion': 'mae', 'max_features': 50, 'n_estimators': 200} 0.701063463314 fetures: 3128 post150_dayofyear_std:0.022417681457041105  , post150_dayofyear_max:0.008964231866345035  , post450_week_25%:0.008679449276035188  , post420_dayofyear_std:0.008378380016366024  , post510_dayofyear_75%:0.008085424156377726  , post360_dayofyear_50%:0.008024360373351991  , post240_dayofyear_50%:0.007976264164219366  , post270_dayofyear_std:0.007899533878504372  , post180_dayofyear_std:0.00789909176487151  , post450_week_std:0.0076043298466074724  , post300_dayofyear_max:0.007592154036674166  , post480_weekofyear_75%:0.00745668852634958  , post390_week_std:0.007327618894522977  , post240_dayofyear_std:0.007305236062653214  , post450_weekofyear_std:0.007295992873653055  , post240_dayofyear_75%:0.0069641583407138745  , post270_dayofyear_50%:0.006866282241384245  , post450_weekofyear_min:0.006762876854847269  , post510_dayofyear_25%:0.006737314430020437  , post330_dayofyear_50%:0.00672964726717036 
16 | 2017-06-11 08:08:48.972957 result: (0.9819418875392917, 0.95461724491246847, 0.65588617424282136, 0.78595625077613818, 0.36854251069819643) params: {'max_features': 50, 'n_estimators': 200, 'criterion': 'mae'} 0.657817091026 fetures: 3128 post150_dayofyear_std:0.020370902095754772  , post150_dayofyear_75%:0.010682919534475801  , post150_dayofyear_mean:0.010356030392783464  , post480_week_25%:0.008065823516717563  , post240_dayofyear_std:0.007880957768625314  , post150_dayofyear_50%:0.007759140214906504  , post450_week_25%:0.007694234383072841  , post480_weekofyear_75%:0.007559134348312591  , post270_dayofyear_max:0.007480062816737922  , post180_dayofyear_std:0.007426108808050903  , post300_dayofyear_max:0.007216634941450267  , post480_weekofyear_max:0.007051269818303007  , post150_dayofyear_25%:0.00673421691242628  , post390_dayofyear_25%:0.0067163490812291735  , post510_dayofyear_std:0.006716206210223072  , post150_dayofyear_max:0.006430728802393112  , post420_dayofyear_std:0.006333905057578368  , post240_dayofyear_max:0.006317673950045931  , post510_dayofyear_25%:0.006245357615756679  , post360_week_std:0.006192215862531334 
17 | 2017-06-11 08:26:04.393517 result: 38.33806318% (0.9826145715111593, 0.95718188839586338, 0.69169719365803672, 0.84069560611894567, 0.41273760426445294) params: {'max_features': 10, 'criterion': 'mae', 'n_estimators': 200} 0.676267991545 fetures: 3128 post420_dayofyear_std:0.008497464164185122  , post240_dayofyear_std:0.007893754602188318  , post480_week_50%:0.007858500539276064  , post180_dayofyear_std:0.0074319413130420385  , post300_dayofyear_std:0.007418788393584342  , post180_dayofyear_max:0.00711614392587529  , post150_dayofyear_std:0.006997454804338879  , post360_dayofyear_std:0.006809163610892495  , post360_week_75%:0.006778430863722091  , post270_week_max:0.006740857018045632  , post360_dayofyear_50%:0.006577623890052819  , post450_dayofyear_max:0.006559779559263823  , post390_dayofyear_75%:0.006499069222878901  , post510_dayofyear_std:0.0064129174153285  , post480_dayofyear_max:0.0062913998389055445  , post510_dayofyear_max:0.00619970746089251  , post360_dayofyear_mean:0.006107403142955725  , post390_dayofyear_50%:0.00603185050789311  , post510_dayofyear_50%:0.006027622495142437  , post210_dayofyear_std:0.005946079297951164 
18 | 2017-06-11 08:47:43.883578 result: (0.9823636978086018, 0.95489379696615817, 0.62649701279962344, 0.85489533006525442, 0.30121598149268813) params: {'criterion': 'mae', 'n_estimators': 100, 'max_features': 10} 0.675404347387 fetures: 3128 post240_dayofyear_std:0.009192095885714542  , post390_dayofyear_50%:0.008651182677963255  , post480_week_50%:0.00855821041340417  , post420_dayofyear_std:0.007845265040992835  , post180_dayofyear_std:0.0077120755814341305  , post420_dayofyear_50%:0.007344936201838517  , post210_dayofyear_std:0.0073311293971738455  , post510_week_mean:0.007197798719444195  , post480_dayofyear_max:0.007164725263317988  , post330_dayofyear_50%:0.007066969554275121  , post180_dayofyear_mean:0.0069429905206699625  , post120_dayofyear_50%:0.006831209231015566  , post390_dayofyear_std:0.00675543155766053  , post240_weekofyear_mean:0.006648836744704547  , post360_dayofyear_std:0.006551403352791161  , post210_dayofyear_50%:0.006418894936123176  , post240_dayofyear_mean:0.006339732474432532  , post300_dayofyear_std:0.006286731820013665  , post240_dayofyear_50%:0.00625809663649062  , post510_weekofyear_count:0.006237365281067994 
19 | 2017-06-11 08:58:51.649481 result: (0.9817621367190471, 0.95666966094253481, 0.62237768632453883, 0.83980002881387417, 0.35138731659108202) params: {'criterion': 'mae', 'n_estimators': 100, 'max_features': 100} 0.656882890725 fetures: 3128 post150_dayofyear_std:0.015542910685266946  , post150_dayofyear_50%:0.014759982697243832  , post150_dayofyear_mean:0.013545261201333545  , post150_dayofyear_25%:0.01236966761255613  , post150_dayofyear_max:0.011920579268849587  , post420_week_25%:0.010954559139424938  , post180_dayofyear_25%:0.010497432544783213  , post150_dayofyear_75%:0.00970027755521497  , post360_dayofyear_max:0.009518843644578893  , post480_dayofyear_25%:0.009043584689071894  , post480_week_mean:0.00882494568153439  , post450_week_25%:0.008615583996591175  , post210_dayofyear_25%:0.008045482435829329  , post360_dayofyear_std:0.007964959618029484  , post330_dayofyear_max:0.007836547041113145  , post510_dayofyear_mean:0.007791970054728534  , post390_dayofyear_25%:0.007726349346811068  , post480_week_25%:0.007699526984838296  , post390_weekofyear_75%:0.007561403892430957  , post510_week_25%:0.007401438532174857 
20 | 2017-06-11 15:58:19.501731 result: (0.9814349581689806, 0.95576951264114907, 0.61674049244621287, 0.73368932669434939, 0.5102332296929849) params: {'max_features': 50, 'criterion': 'mae', 'n_estimators': 100} 0.629054129826 fetures: 3128 post150_dayofyear_mean:0.014807172835698192  , post150_dayofyear_50%:0.013605929680389157  , post510_dayofyear_std:0.011998228766294779  , post480_week_mean:0.01104928578506366  , post240_dayofyear_std:0.010920012257339112  , post150_dayofyear_std:0.010487458558428324  , post450_week_25%:0.010252480949790194  , post150_dayofyear_25%:0.009360222139004311  , post420_week_25%:0.00798410875661814  , post150_dayofyear_75%:0.007869174735869134  , post510_dayofyear_count:0.007721122815259382  , post180_dayofyear_std:0.007646612759851348  , post510_dayofyear_25%:0.007413240322320549  , post270_dayofyear_max:0.007378546552945636  , post420_dayofyear_std:0.007249181136462292  , post240_dayofyear_max:0.007226312648989882  , post480_dayofyear_25%:0.007070355456462329  , post270_dayofyear_25%:0.006909248872775168  , post420_weekofyear_75%:0.006903978073348  , post390_week_count:0.006882762448960537 
21 | 2017-06-11 16:33:37.162570 result: (0.9816843324397545, 0.90515435921606147, 0.34905165144414896, 0.4671767465679193, 0.038083361996820608) params: {'n_estimators': 100, 'criterion': 'mae', 'max_features': 'sqrt'} 0.425271058438 fetures: 2576 post240_weather_max_std:0.012864649575464859  , post270_weekofyear_std:0.0126198860921078  , post360_weather_max_max:0.010935212787435826  , post30_weather_min_std:0.010769899381306085  , post390_weather_max_std:0.010519958358186351  , post300_weather_max_max:0.010432196502535225  , post390_weather_max_max:0.010428378496249273  , post270_week_std:0.01030731990801671  , post150_weekofyear_75%:0.009745472574313243  , post360_week_max:0.009327295806867949  , post330_weather_max_std:0.009080110129851581  , post120_weather_max_25%:0.009001006770954039  , post120_weather_max_min:0.00879535211580983  , post150_weather_max_count:0.00877089502845152  , post120_weekofyear_min:0.008741762775095698  , post360_weekofyear_max:0.008721351251843831  , post300_week_std:0.008613261447946932  , post420_weather_min_max:0.008610996373947242  , post30_weather_min_max:0.008345216958299066  , post210_weekofyear_std:0.008289296068348771 
22 | 2017-06-11 16:40:47.595154 result: (0.9810588480757253, 0.9009934953760721, 0.47064635514010245, 0.77357657917553135, 0.079661131009523656) params: {'max_features': 10, 'criterion': 'mae', 'n_estimators': 100} 0.486083862193 fetures: 2576 post150_weekofyear_mean:0.009855215160781446  , post270_week_max:0.009424304452172745  , post150_weather_max_std:0.009179089897462033  , post390_weather_max_std:0.00882516862532233  , post270_week_std:0.008645072333986885  , post30_weather_min_mean:0.008559025833251774  , post120_weather_max_25%:0.008226272369590993  , post240_weather_max_mean:0.00744704933762555  , post30_weather_min_max:0.0070183445186746615  , post180_week_max:0.0069845468590864004  , post150_weather_min_count:0.006816804954875264  , post210_weather_min_max:0.006768159980661166  , post330_week_25%:0.006710496086316952  , post360_week_std:0.006707748614139498  , post30_weather_min_count:0.006610109615358368  , post360_dayofyear_max:0.006600819659005956  , post420_week_max:0.0065661221188512385  , post390_week_max:0.006525288637539832  , post330_weekofyear_mean:0.006243911853077088  , post210_week_max:0.006066934639645646 
23 | 2017-06-11 19:11:56.325843 result: (0.9784196521092927, 0.94554580196881099, 0.53083376740955912, 0.76560963212100075, 0.23026712372690553) params: {'n_estimators': 100} 0.595904515492 fetures: 947 post30_dayofyear_25%:0.11811921791641394  , post30_dayofyear_count:0.07579042696019553  , post90_weather_max_std:0.056591444366807  , post30_dayofyear_std:0.046026138026924256  , post60_weather_max_std:0.04162110461211387  , post30_dayofyear_mean:0.025382906526467672  , post180_dayofweek_50%:0.022430012564422307  , post180_dayofyear_mean_consumption:0.021604213790490507  , post360_dayofyear_mean_consumption:0.018768060346445497  , post30_dayofyear_50%:0.017633352968677737  , post30_dayofyear_max:0.01680462663812277  , post30_dayofyear_mean_consumption:0.015237032570365959  , post360_weather_min_mean:0.014943015954353099  , post60_weather_max_75%:0.014212583959111916  , post360_weather_min_std:0.012485563991450574  , post60_dayofyear_mean_consumption:0.011525453469041776  , post180_dayofyear_25%:0.010847699219107021  , post180_weather_max_75%:0.010164706192497814  , post90_weather_max_75%:0.009370928423593057  , post90_weather_min_std:0.00932922385884032 
24 | 2017-06-11 19:29:49.236588 result: (0.9785965129060097, 0.94653241011561007, 0.53755270245661246, 0.76393584791510372, 0.25302397234282148) params: {'criterion': 'mse', 'n_estimators': 300} 0.596969059163 fetures: 947 post30_dayofyear_25%:0.09167544904606786  , post30_dayofyear_count:0.08591894356458817  , post90_weather_max_std:0.06076611161894268  , post30_dayofyear_std:0.046122230531716935  , post60_weather_max_std:0.03888173093931677  , post30_dayofyear_max:0.031669988689992434  , post180_dayofweek_50%:0.02399328462044762  , post180_dayofyear_mean_consumption:0.020319509482770916  , post30_dayofyear_50%:0.017601774067209627  , post30_dayofyear_mean:0.017218235631364813  , post90_dayofyear_mean_consumption:0.015498407303676718  , post30_dayofyear_mean_consumption:0.014988022017633964  , post360_weather_min_mean:0.014787352206729346  , post360_weather_min_std:0.013178957455030355  , post60_weather_max_75%:0.013066742045006054  , post60_dayofyear_mean_consumption:0.012579454319031442  , post180_dayofyear_25%:0.01194076805497424  , post360_dayofyear_mean_consumption:0.011908989472083768  , post180_weather_max_75%:0.011077881804171468  , post90_weather_min_std:0.011029403488666262 
25 | 2017-06-11 20:04:12.628598 result: (0.979350828168523, 0.94996825171369259, 0.52923490324997391, 0.76152951699607996, 0.22204518053803912) params: {'max_features': 100, 'n_estimators': 500} 0.545932166442 fetures: 947 post30_dayofyear_mean:0.03292709069686641  , post30_dayofyear_mean_consumption:0.029643831644066528  , post360_dayofyear_mean_consumption:0.02770423732034277  , post30_dayofyear_std:0.027549057148796817  , post180_dayofyear_mean_consumption:0.027093311999831674  , post60_dayofyear_mean_consumption:0.026930525487501192  , post30_dayofyear_count:0.025774574886522947  , post30_dayofyear_75%:0.025636980550907962  , post30_dayofyear_max:0.023392608001917482  , post30_dayofyear_25%:0.022398460960013456  , post30_dayofyear_50%:0.02203899787249782  , post90_weather_max_std:0.02042789461889396  , post60_weather_max_std:0.019837906795038853  , post90_dayofyear_mean_consumption:0.017000500001235756  , post30_dayofyear_min:0.01588341954651439  , post60_weather_max_max:0.014604075074070603  , post90_weather_max_max:0.010931682451005266  , post90_weather_min_max:0.010719345903978023  , post90_weather_max_75%:0.009568806874345834  , post90_weather_min_std:0.009284540257766132 
26 | 2017-06-11 20:52:11.154737 result: (0.9805667292777991, 0.9528420677391769, 0.51928078952621559, 0.80357155376042555, 0.098488845197916675) params: {'max_features': 100, 'n_estimators': 100} 0.61073342904 fetures: 947 post360_dayofyear_mean_consumption:0.03704127417131781  , post30_dayofyear_mean:0.034040630835957056  , post30_dayofyear_mean_consumption:0.03210342153072227  , post180_dayofyear_mean_consumption:0.03208407181088475  , post60_dayofyear_mean_consumption:0.028872098095710272  , post90_dayofyear_mean_consumption:0.024996829126474912  , post180_dayofyear_min:0.023715415253222343  , post30_dayofyear_25%:0.02272970173784293  , post180_dayofyear_count:0.02254698798726697  , post90_dayofyear_min:0.02062065977948385  , post30_dayofyear_std:0.020078511839485735  , post30_dayofyear_75%:0.019841721598996668  , post60_dayofyear_min:0.015718063029756937  , post30_dayofyear_min:0.015673031703683372  , post60_weather_max_mean:0.014934283181800512  , post360_weather_max_max:0.014130325246507858  , post360_dayofyear_count:0.013041758458654261  , post60_weather_max_std:0.012432604903157847  , post60_weather_max_50%:0.011893318693613278  , post30_dayofyear_50%:0.010682019044875845 
27 | 2017-06-11 21:14:08.918051 result: (0.9796536193377529, 0.94988101646402312, 0.49657570196083539, 0.80299260406147277, 0.14774362853862943) params: {'n_estimators': 300, 'max_features': 100} 0.568236803016 fetures: 947 post30_dayofyear_mean:0.04497040334629134  , post60_dayofyear_mean_consumption:0.03916903161800868  , post30_dayofyear_mean_consumption:0.031710198517904555  , post180_dayofyear_mean_consumption:0.031109698338844995  , post360_dayofyear_mean_consumption:0.029934374947765262  , post30_dayofyear_75%:0.028947422563007745  , post90_dayofyear_mean_consumption:0.024697913691739386  , post30_dayofyear_25%:0.02260825137530281  , post30_dayofyear_std:0.021892750568281452  , post30_dayofyear_max:0.018446171965005188  , post90_weather_max_std:0.015626536232576142  , post30_dayofyear_count:0.015376368397690627  , post360_weather_max_max:0.014511203842828098  , post30_dayofyear_50%:0.01401841199589714  , post360_weather_max_std:0.013878619442006015  , post30_dayofyear_min:0.012756048357801452  , post90_weather_max_25%:0.01268616867979231  , post90_weather_max_max:0.012251838601502649  , post360_weather_max_25%:0.012127789019530086  , post60_weather_max_max:0.01134733090872791 
28 | 2017-06-11 22:05:28.309707 result: (0.9860422278222353, 0.97538373196296657, 0.72281602842497583, 0.90777480934263699, 0.46913574740101072) params: {'n_estimators': 200, 'max_features': 100} 0.757100555647 fetures: 947 post360_is_festival_mean:0.051585150706398394  , post30_is_week_std:0.02815793070171819  , post30_is_holiday_std:0.027995437386961308  , post90_dayofyear_mean_consumption:0.025973191190086443  , post180_dayofyear_mean_consumption:0.02447763976522031  , post360_dayofyear_mean_consumption:0.0171762447855321  , post30_dayofyear_mean_consumption:0.0165617283461739  , post60_is_week_std:0.01515694397368341  , post90_is_month_end_std:0.011801336931753064  , post360_is_festival_75%:0.011193087365073111  , post90_is_year_start_max:0.010451485991394378  , post60_is_year_start_max:0.010268371091034042  , post60_dayofyear_mean_consumption:0.009455856333778252  , post60_is_festival_std:0.009032913608421537  , post30_is_weekend_std:0.00892505027556921  , post360_is_festival_50%:0.008745003892332374  , post30_dayofweek_std:0.0085378125824328  , post30_is_year_start_std:0.00849014511263141  , post360_weekofyear_std:0.00841137041836314  , post360_is_year_start_count:0.008197045563485762 
29 | 2017-06-11 23:26:37.041631 result: (0.9859796371561408, 0.9754629518756025, 0.71896196585158012, 0.90887844280777552, 0.48842980444506767) params: {'max_features': 100, 'n_estimators': 500} 0.743003458272 fetures: 882 post366_is_festival_mean:0.04721796907113417  , post366_is_festival_count:0.036851157336773376  , post30_is_holiday_std:0.03639024586547476  , post360_dayofyear_mean_consumption:0.02582299981241632  , post30_is_week_std:0.025110922405546087  , post60_dayofyear_mean_consumption:0.02067480042687998  , post180_dayofyear_mean_consumption:0.017889890325557704  , post90_dayofyear_mean_consumption:0.017332422986329362  , post30_dayofyear_mean_consumption:0.016831391647010974  , post366_week_std:0.01072675522161596  , post60_is_festival_25%:0.010072623362054013  , post366_is_month_start_mean:0.009655833457309049  , post90_is_year_start_max:0.009624668374092649  , post90_is_quarter_start_std:0.009355447197347547  , post60_is_holiday_std:0.008616167666026213  , post30_dayofweek_std:0.008457193856671786  , post366_is_year_start_mean:0.008360823934299535  , post366_weekofyear_std:0.007953899430910173  , post30_is_year_start_std:0.006969789588555151  , post180_is_year_start_std:0.006852640492247664 
30 | 2017-06-11 23:51:24.220140 result: (0.9860098827620267, 0.97493707538390306, 0.69982191346224276, 0.89222561614429197, 0.41652547120700678) params: {'max_features': 100, 'n_estimators': 500} 0.744737662618 fetures: 882 post366_is_festival_mean:0.04785729282430978  , post366_is_festival_count:0.037254644835714716  , post30_is_holiday_std:0.035718516225495024  , post30_is_week_std:0.02917860355956139  , post90_is_year_start_max:0.014104182303085307  , post366_is_year_start_count:0.011621301926753383  , post60_is_year_start_max:0.009972666493059302  , post90_is_month_start_std:0.009771885074172862  , post366_is_month_start_mean:0.009694197503834653  , post60_is_festival_25%:0.009659546055855069  , post180_is_month_end_std:0.009375579410585722  , post30_dayofweek_std:0.008812767342725448  , post90_is_month_end_std:0.00868900422020309  , post60_is_holiday_std:0.008605141220977287  , post30_dayofyear_mean_consumption:0.007769569461819031  , post60_is_week_std:0.007420453968558346  , post30_is_weekend_std:0.006848387056812638  , post366_dayofyear_std:0.006844018626798128  , post90_dayofyear_mean_consumption:0.006075090620651104  , post180_weather_min_50%:0.005969549182348032 
31 | 2017-06-12 00:05:43.794702 result: (0.9860706488665998, 0.97534921354359272, 0.72078167119037273, 0.90363782784957447, 0.48863810474133329) params: {'n_estimators': 200, 'max_features': 50} 0.743158037935 fetures: 882 post366_is_festival_mean:0.03649376299105978  , post30_is_holiday_std:0.02258167887820268  , post30_is_week_std:0.022044272893742566  , post360_dayofyear_mean_consumption:0.019963456957026712  , post366_is_festival_count:0.01882470095489192  , post90_is_year_start_max:0.01633665204149714  , post180_dayofyear_mean_consumption:0.01567402044806286  , post60_dayofyear_mean_consumption:0.013673382475315347  , post60_is_festival_25%:0.010670982718640002  , post30_dayofyear_mean_consumption:0.010559358327522162  , post366_is_month_start_mean:0.009480563553280058  , post60_wind_type_75%:0.009216229588183326  , post90_dayofyear_mean_consumption:0.008846540951297997  , post366_week_std:0.008812853971518767  , post30_is_weekend_std:0.008470241561017753  , post60_is_week_std:0.008188887934216849  , post60_is_holiday_std:0.007627849145458359  , post366_is_month_end_mean:0.007337613834769441  , post60_is_festival_std:0.007012774808170126  , post366_dayofyear_std:0.006611882995252664 
32 | 2017-06-12 07:23:35.557214 result: 63.98888347% (0.9866254384003387, 0.97799788949857136, 0.73936076203923151, 0.91569187187148926, 0.51828353005522221) params: {'max_features': 50, 'n_estimators': 500} 0.777988742549 fetures: 670 post30_is_holiday_std:0.03050903718051737  , post366_is_festival_count:0.02943960513768783  , post366_is_festival_mean:0.02725363727286884  , post30_is_week_std:0.020258481676271825  , post60_dayofyear_mean_consumption:0.016901125033073568  , post30_dayofyear_mean_consumption:0.015474243304829525  , post180_dayofyear_mean_consumption:0.015311166084897654  , post90_dayofyear_mean_consumption:0.014101101716247481  , post360_dayofyear_mean_consumption:0.01243032019748404  , post30_is_weekend_std:0.012062861295601066  , post60_wind_type_75%:0.011108343014274045  , post30_dayofweek_std:0.010988796545960702  , post60_is_week_std:0.010505236412668078  , post366_week_std:0.009535675996744628  , dayofyear:0.009120641924424624  , post366_days_in_month_25%:0.00859323859809345  , weather_min:0.008308283289756335  , post366_is_festival_50%:0.007882745783252047  , post60_is_holiday_std:0.007720773225235407  , post180_days_in_month_25%:0.007471641316633146 
33 | 2017-06-12 07:55:57.083013 result: (0.98647077082579, 0.97766446838697241, 0.73736552870742533, 0.91550905640307811, 0.51525948359136309) params: {'n_estimators': 500, 'max_features': 200} 0.770428631298 fetures: 1150 post366_is_festival_count:0.0646681767325599  , post366_is_festival_mean:0.056423693956536305  , post30_is_week_std:0.0455505117833928  , post30_is_holiday_std:0.04338312404183893  , post180_dayofyear_mean_consumption:0.02487604783294715  , post90_dayofyear_mean_consumption:0.02452895920170585  , post360_dayofyear_mean_consumption:0.02340830246292294  , post60_dayofyear_mean_consumption:0.02097400094658198  , post30_dayofyear_mean_consumption:0.0194873419557608  , post60_is_holiday_std:0.01278209594108359  , dayofyear:0.012252390059380178  , post366_week_std:0.012062631412360652  , post30_dayofweek_std:0.009921234681556349  , post366_weekofyear_std:0.009385565374136451  , post60_is_week_std:0.008502364159111568  , post30_is_weekend_std:0.008131421733937983  , post300_is_festival_50%:0.007739547507486541  , post300_is_weekend_count:0.007084020430866262  , week:0.006933822446656162  , weather_min:0.00680273392220532 
34 | 2017-06-12 08:29:16.440120 result: (0.9857697957310936, 0.97520264450998351, 0.74766783959860228, 0.91986322339539617, 0.56379069236105284) params: {'max_features': 100, 'n_estimators': 100} 0.783397061484 fetures: 710 post366_is_festival_is0_day_consumption:0.053009558125623926  , post366_is_festival_mean:0.045808018906376535  , post30_is_week_std:0.04033248224727701  , post366_is_festival_count:0.032184321358420726  , post180_dayofyear_mean_consumption:0.0312250898045562  , post30_is_holiday_std:0.03054884518165129  , post90_dayofyear_mean_consumption:0.029286292769630674  , post60_wind_type_75%:0.02402238487725719  , post360_dayofyear_mean_consumption:0.022255195079653192  , post30_dayofyear_mean_consumption:0.020338018451951868  , post60_dayofyear_mean_consumption:0.01671504236551689  , dayofyear:0.012799754178785783  , post180_days_in_month_mean:0.012572628428667238  , post60_is_holiday_std:0.01151770828347939  , post30_weather_type_std:0.01036604896398658  , week:0.009121525707887912  , post366_month_mean:0.008995026879686969  , post366_is_festival_75%:0.008788192227401133  , weather_min:0.00873260683030331  , post366_weather_max_count:0.00808530625765832 
35 | 2017-06-12 19:14:40.093735 result: (0.9862574473283529, 0.97567068407074931, 0.7350667487545588, 0.89718025745485819, 0.53555567174279206) params: {'max_features': 100, 'n_estimators': 100} 0.771231894296 fetures: 979 post366_is_festival_count:0.05078995930825454  , post366_is_festival_25%_day_consumption:0.03655333121927563  , post366_is_festival_count_day_consumption:0.034339266133819775  , post30_is_week_std:0.03409284967110627  , post366_is_festival_mean_day_consumption:0.027057561605615767  , post366_is_festival_mean:0.024542855052014607  , post366_is_week_std_day_consumption:0.020112807358730535  , post366_is_holiday_std_day_consumption:0.01903976405461335  , post366_is_festival_50%_day_consumption:0.018864031016925794  , post366_is_week_min_day_consumption:0.01527043640493849  , post30_is_holiday_std:0.014907330801875033  , post366_days_in_month_max:0.01289563882751284  , post366_is_holiday_min_day_consumption:0.012335367334325144  , dayofyear:0.01161472135940337  , post366_week_max_day_consumption:0.010497489761341829  , post180_is_festival_mean:0.008668883992996224  , post366_weekofyear_std:0.008658860890646651  , post180_is_festival_50%_day_consumption:0.00865598854773328  , post366_week_std:0.008076164665792144  , post366_month_max_day_consumption:0.007685215631141424 
36 | 2017-06-12 19:25:19.073365 result: (0.9867991261621545, 0.97766777023452556, 0.7367590881109578, 0.90249629847624457, 0.56241398970882805) params: {'max_features': 100, 'n_estimators': 200} 0.770315820758 fetures: 979 post366_is_festival_count:0.040799545234957754  , post366_is_festival_mean_day_consumption:0.0395232336344848  , post366_is_festival_25%_day_consumption:0.03750448172308711  , post366_is_festival_count_day_consumption:0.0352292451957175  , post366_is_festival_mean:0.03307791110328274  , post366_is_festival_50%_day_consumption:0.027910759969119275  , post30_is_week_std:0.023358584356657115  , post30_is_holiday_std:0.018964289843085864  , post366_is_week_std_day_consumption:0.018350469580544503  , post366_is_holiday_std_day_consumption:0.016713986637495352  , post366_is_week_min_day_consumption:0.015499683260332877  , dayofyear:0.009854263631475344  , post366_is_holiday_min_day_consumption:0.009259430501178311  , post366_week_std:0.008391921034101532  , week:0.008333252422592408  , post366_week_max_day_consumption:0.008150295848354715  , post30_dayofweek_std:0.008086278138478816  , post180_is_festival_mean:0.00790448004931459  , post90_is_holiday_std_day_consumption:0.007325533253923696  , post366_weekofyear_std:0.007046368021324972 
37 | 2017-06-12 20:50:22.306037 result: (0.9860635178227071, 0.97548880480927813, 0.73580500529532256, 0.89254671765264848, 0.57712359121888446) params: {'n_estimators': 100, 'max_features': 100} 0.748240129169 fetures: 6315 post366_post366_combine_quarter_and_is_festival_50%_day_consumption:0.02221312568471655  , post366_post366_combine_week_and_is_festival_mean_day_consumption:0.017795924159881484  , post366_post366_combine_week_and_is_festival_min_day_consumption:0.013661559410389981  , post90_post90_combine_week_and_is_festival_min_day_consumption:0.013370064655104154  , post366_post366_combine_dayofyear_and_is_festival_50%_day_consumption:0.012841577310095045  , post366_post366_combine_days_in_month_and_is_festival_mean_day_consumption:0.011008253032063627  , post366_post366_combine_year_and_is_festival_25%_day_consumption:0.009871354645920934  , post366_post366_combine_week_and_is_festival_25%_day_consumption:0.009276392455392397  , post366_post366_combine_weather_max_and_is_festival_50%_day_consumption:0.008991516025441473  , post366_post366_combine_weather_max_and_is_festival_25%_day_consumption:0.00892902318983094  , post366_post366_combine_weather_max_and_is_festival_count_day_consumption:0.008825670661912508  , post366_post366_combine_weekofyear_and_is_festival_count_day_consumption:0.008584675661534142  , post366_post366_combine_month_and_is_festival_50%_day_consumption:0.008467547043043849  , post30_is_week_std:0.008214512984556981  , post366_is_festival_mean_day_consumption:0.008140061461525879  , post366_post366_combine_days_in_month_and_is_week_50%_day_consumption:0.007978852351534515  , post366_post366_combine_month_and_is_festival_std_day_consumption:0.007531485636624336  , post366_post366_combine_year_and_is_festival_mean_day_consumption:0.007517912583840054  , post366_post366_combine_month_and_is_festival_25%_day_consumption:0.0073852186900209395  , post30_post30_combine_quarter_and_rare_weather_std_day_consumption:0.00628694275045872 
38 | 2017-06-12 21:07:18.433023 result: (0.9875340316757467, 0.97970303308119733, 0.75657589321922436, 0.91571876988603362, 0.59606956020233115) params: {'max_features': 100, 'n_estimators': 500} 0.774661486377 fetures: 1666 post366_post366_combine_days_in_month_and_is_festival_count_day_consumption:0.013616842976884605  , post366_is_festival_25%_day_consumption:0.013511528563406282  , post366_post366_combine_days_in_month_and_is_festival_50%_day_consumption:0.013298023062819792  , post366_post366_combine_quarter_and_is_festival_50%_day_consumption:0.012523627379392574  , post180_post180_combine_week_and_is_festival_min_day_consumption:0.012367097148214145  , post366_post366_combine_dayofyear_and_is_festival_25%_day_consumption:0.01232556498664281  , post366_post366_combine_weekofyear_and_is_festival_mean_day_consumption:0.011291173850269115  , post90_post90_combine_weekofyear_and_is_festival_min_day_consumption:0.01099491105470402  , post366_post366_combine_weather_min_and_is_festival_25%_day_consumption:0.010886237130079735  , post366_post366_combine_week_and_is_festival_count_day_consumption:0.010294204094960957  , post366_is_festival_mean:0.010237850644440431  , post366_post366_combine_month_and_is_festival_25%_day_consumption:0.009964728734108365  , post366_post366_combine_week_and_is_festival_25%_day_consumption:0.009464475947519575  , post366_post366_combine_year_and_is_festival_count_day_consumption:0.009410767547369867  , post366_post366_combine_month_and_is_festival_50%_day_consumption:0.009405389469942706  , post366_post366_combine_weather_min_and_is_festival_mean_day_consumption:0.009191534959431901  , post366_post366_combine_dayofyear_and_is_festival_mean_day_consumption:0.008877472914973187  , post366_post366_combine_days_in_month_and_is_festival_mean_day_consumption:0.008677306241352722  , post366_is_festival_50%_day_consumption:0.008439029435670908  , post366_post366_combine_weather_max_and_is_festival_count_day_consumption:0.00821951136887792 
39 | 2017-06-12 22:23:55.298309 result: (0.9868607500233955, 0.97798381605676776, 0.74116781023157952, 0.89751405429135322, 0.53704631521278068) params: {'n_estimators': 200, 'max_features': 200} 0.768118651739 fetures: 3000 post366_post366_combine_weekofyear_and_is_festival_mean_day_consumption:0.014474165442483193  , post366_post366_combine_weekofyear_and_is_festival_25%_day_consumption:0.012691197563604437  , post366_post366_combine_days_in_month_and_is_festival_50%_day_consumption:0.012617939859521598  , post366_post366_combine_days_in_month_and_wind_type_75%_day_consumption:0.011543684688620787  , post366_is_festival_count:0.011124442902471105  , post180_post180_combine_days_in_month_and_wind_type_min_day_consumption:0.010662163622520378  , post366_post366_combine_year_and_is_festival_50%_day_consumption:0.010417934960333995  , post366_post366_combine_quarter_and_is_festival_mean_day_consumption:0.010211410961749379  , post90_post90_combine_weekofyear_and_is_festival_min_day_consumption:0.010075932804518472  , post366_post366_combine_days_in_month_and_is_holiday_std_day_consumption:0.009947585288546551  , post366_post366_combine_weather_max_and_is_week_std_day_consumption:0.009581154633333824  , post366_post366_combine_dayofyear_and_is_festival_50%_day_consumption:0.009443225013791821  , post366_post366_combine_month_and_is_festival_mean_day_consumption:0.009348100429311995  , post30_is_week_std:0.009092924828265432  , post366_post366_combine_year_and_is_festival_mean_day_consumption:0.008965047271731115  , post90_post90_combine_week_and_is_festival_min_day_consumption:0.008933227794773128  , post366_is_festival_50%_day_consumption:0.008656116175382443  , post180_post180_combine_weekofyear_and_is_festival_min_day_consumption:0.008619167153293968  , post366_post366_combine_week_and_is_festival_25%_day_consumption:0.008598785575571626  , post180_post180_combine_dayofyear_and_is_festival_min_day_consumption:0.008444706483637206 
40 | 2017-06-12 23:29:52.065789 result:
41 | 2017-06-12 23:45:41.635635 result: (1.0, 1.0, 0.74049079281277497, 0.83801583802054425, 0.66110825830568487)
42 | 2017-06-13 06:36:07.164962 result: (0.9999999999999999, 1.0, 0.74169914680087734, 0.84048503686220044, 0.6658133990544437)
43 | 2017-06-13 07:55:11.229020 result: (0.9876500688451734, 0.98078256144556675, 0.78898896165565757, 0.86008654041103993, 0.68182410751042843)
44 | 


--------------------------------------------------------------------------------
/code/rules.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 1,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "from datetime import datetime, timedelta\n",
 12 |     "\n",
 13 |     "import numpy as np\n",
 14 |     "import pandas as pd\n",
 15 |     "import scipy as sp"
 16 |    ]
 17 |   },
 18 |   {
 19 |    "cell_type": "code",
 20 |    "execution_count": 2,
 21 |    "metadata": {
 22 |     "collapsed": true
 23 |    },
 24 |    "outputs": [],
 25 |    "source": [
 26 |     "indexs = pd.date_range('2016/10/1', '2016/10/31')"
 27 |    ]
 28 |   },
 29 |   {
 30 |    "cell_type": "code",
 31 |    "execution_count": 3,
 32 |    "metadata": {
 33 |     "collapsed": true
 34 |    },
 35 |    "outputs": [],
 36 |    "source": [
 37 |     "def gen_df(g):\n",
 38 |     "    df = pd.DataFrame(index=indexs, columns=['power_consumption'])\n",
 39 |     "    s = pd.Series(g.power_consumption, index=indexs)\n",
 40 |     "    df['power_consumption'] = s\n",
 41 |     "    df.fillna(df.power_consumption.mean(), inplace=True)\n",
 42 |     "    return df"
 43 |    ]
 44 |   },
 45 |   {
 46 |    "cell_type": "code",
 47 |    "execution_count": 4,
 48 |    "metadata": {
 49 |     "collapsed": true
 50 |    },
 51 |    "outputs": [],
 52 |    "source": [
 53 |     "train_path = '../dataset/train.csv'"
 54 |    ]
 55 |   },
 56 |   {
 57 |    "cell_type": "code",
 58 |    "execution_count": 5,
 59 |    "metadata": {
 60 |     "collapsed": true
 61 |    },
 62 |    "outputs": [],
 63 |    "source": [
 64 |     "train_df = pd.read_csv(train_path, parse_dates=['record_date'])"
 65 |    ]
 66 |   },
 67 |   {
 68 |    "cell_type": "code",
 69 |    "execution_count": 6,
 70 |    "metadata": {
 71 |     "collapsed": false
 72 |    },
 73 |    "outputs": [],
 74 |    "source": [
 75 |     "train_df['record_date'] = train_df.record_date + timedelta(days=31)"
 76 |    ]
 77 |   },
 78 |   {
 79 |    "cell_type": "code",
 80 |    "execution_count": 8,
 81 |    "metadata": {
 82 |     "collapsed": false
 83 |    },
 84 |    "outputs": [],
 85 |    "source": [
 86 |     "result_df = train_df.set_index('record_date').groupby(['user_id']).apply(gen_df)"
 87 |    ]
 88 |   },
 89 |   {
 90 |    "cell_type": "code",
 91 |    "execution_count": 9,
 92 |    "metadata": {
 93 |     "collapsed": false
 94 |    },
 95 |    "outputs": [],
 96 |    "source": [
 97 |     "submit_df = result_df.reset_index().groupby('level_1')[['power_consumption']].sum()"
 98 |    ]
 99 |   },
100 |   {
101 |    "cell_type": "code",
102 |    "execution_count": 10,
103 |    "metadata": {
104 |     "collapsed": false
105 |    },
106 |    "outputs": [],
107 |    "source": [
108 |     "submit_df['predict_date'] = submit_df.index.map(lambda x:x.strftime('%Y%m%d'))"
109 |    ]
110 |   },
111 |   {
112 |    "cell_type": "code",
113 |    "execution_count": 11,
114 |    "metadata": {
115 |     "collapsed": false
116 |    },
117 |    "outputs": [],
118 |    "source": [
119 |     "submit_df['predict_power_consumption'] = submit_df.power_consumption.astype(int)"
120 |    ]
121 |   },
122 |   {
123 |    "cell_type": "code",
124 |    "execution_count": 12,
125 |    "metadata": {
126 |     "collapsed": false
127 |    },
128 |    "outputs": [],
129 |    "source": [
130 |     "submit_df.to_csv('../Tianchi_power_predict_table.csv', columns=['predict_date', 'predict_power_consumption'], index=False)"
131 |    ]
132 |   }
133 |  ],
134 |  "metadata": {
135 |   "anaconda-cloud": {},
136 |   "hide_input": false,
137 |   "kernelspec": {
138 |    "display_name": "Python 3",
139 |    "language": "python",
140 |    "name": "python3"
141 |   },
142 |   "language_info": {
143 |    "codemirror_mode": {
144 |     "name": "ipython",
145 |     "version": 3
146 |    },
147 |    "file_extension": ".py",
148 |    "mimetype": "text/x-python",
149 |    "name": "python",
150 |    "nbconvert_exporter": "python",
151 |    "pygments_lexer": "ipython3",
152 |    "version": "3.5.1"
153 |   }
154 |  },
155 |  "nbformat": 4,
156 |  "nbformat_minor": 2
157 | }
158 | 


--------------------------------------------------------------------------------
/code/run.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": null,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "!date"
 12 |    ]
 13 |   },
 14 |   {
 15 |    "cell_type": "markdown",
 16 |    "metadata": {},
 17 |    "source": [
 18 |     "# 规则,前一个月平均值 "
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": null,
 24 |    "metadata": {
 25 |     "collapsed": true
 26 |    },
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "extract_ipy = 'trick_extract.ipynb'\n",
 30 |     "extract_py = 'trick_extract.py'"
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": null,
 36 |    "metadata": {
 37 |     "collapsed": false
 38 |    },
 39 |    "outputs": [],
 40 |    "source": [
 41 |     "# !jupyter nbconvert --to python $extract_ipy"
 42 |    ]
 43 |   },
 44 |   {
 45 |    "cell_type": "code",
 46 |    "execution_count": null,
 47 |    "metadata": {
 48 |     "collapsed": false
 49 |    },
 50 |    "outputs": [],
 51 |    "source": [
 52 |     "# !python $extract_py test"
 53 |    ]
 54 |   },
 55 |   {
 56 |    "cell_type": "markdown",
 57 |    "metadata": {},
 58 |    "source": [
 59 |     "## 切分数据"
 60 |    ]
 61 |   },
 62 |   {
 63 |    "cell_type": "code",
 64 |    "execution_count": null,
 65 |    "metadata": {
 66 |     "collapsed": false
 67 |    },
 68 |    "outputs": [],
 69 |    "source": [
 70 |     "!runipy split_samples.ipynb"
 71 |    ]
 72 |   },
 73 |   {
 74 |    "cell_type": "markdown",
 75 |    "metadata": {},
 76 |    "source": [
 77 |     "# 过滤清洗数据"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": null,
 83 |    "metadata": {
 84 |     "collapsed": false
 85 |    },
 86 |    "outputs": [],
 87 |    "source": [
 88 |     "!jupyter nbconvert --to python clean_data.ipynb"
 89 |    ]
 90 |   },
 91 |   {
 92 |    "cell_type": "code",
 93 |    "execution_count": null,
 94 |    "metadata": {
 95 |     "collapsed": false
 96 |    },
 97 |    "outputs": [],
 98 |    "source": [
 99 |     "!python clean_data.py test"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "code",
104 |    "execution_count": null,
105 |    "metadata": {
106 |     "collapsed": false
107 |    },
108 |    "outputs": [],
109 |    "source": [
110 |     "for i in range(1, 10):\n",
111 |     "    !python clean_data.py train $i"
112 |    ]
113 |   },
114 |   {
115 |    "cell_type": "markdown",
116 |    "metadata": {},
117 |    "source": [
118 |     "## 提取特征 "
119 |    ]
120 |   },
121 |   {
122 |    "cell_type": "code",
123 |    "execution_count": null,
124 |    "metadata": {
125 |     "collapsed": true
126 |    },
127 |    "outputs": [],
128 |    "source": [
129 |     "extract_ipy = 'extract_fetures.ipynb'\n",
130 |     "extract_py = 'extract_fetures.py'"
131 |    ]
132 |   },
133 |   {
134 |    "cell_type": "code",
135 |    "execution_count": null,
136 |    "metadata": {
137 |     "collapsed": false
138 |    },
139 |    "outputs": [],
140 |    "source": [
141 |     "!jupyter nbconvert --to python $extract_ipy"
142 |    ]
143 |   },
144 |   {
145 |    "cell_type": "code",
146 |    "execution_count": null,
147 |    "metadata": {
148 |     "collapsed": true
149 |    },
150 |    "outputs": [],
151 |    "source": [
152 |     "for i in range(1, 10):    \n",
153 |     "    !python $extract_py train $i"
154 |    ]
155 |   },
156 |   {
157 |    "cell_type": "code",
158 |    "execution_count": null,
159 |    "metadata": {
160 |     "collapsed": true
161 |    },
162 |    "outputs": [],
163 |    "source": [
164 |     "!python $extract_py test"
165 |    ]
166 |   },
167 |   {
168 |    "cell_type": "code",
169 |    "execution_count": null,
170 |    "metadata": {
171 |     "collapsed": true
172 |    },
173 |    "outputs": [],
174 |    "source": [
175 |     "!runipy combine_fetures.ipynb"
176 |    ]
177 |   },
178 |   {
179 |    "cell_type": "markdown",
180 |    "metadata": {},
181 |    "source": [
182 |     "## 训练模型 "
183 |    ]
184 |   },
185 |   {
186 |    "cell_type": "code",
187 |    "execution_count": null,
188 |    "metadata": {
189 |     "collapsed": false,
190 |     "scrolled": false
191 |    },
192 |    "outputs": [],
193 |    "source": [
194 |     "!runipy train_model.ipynb"
195 |    ]
196 |   },
197 |   {
198 |    "cell_type": "markdown",
199 |    "metadata": {},
200 |    "source": [
201 |     "## 模型预测 "
202 |    ]
203 |   },
204 |   {
205 |    "cell_type": "code",
206 |    "execution_count": null,
207 |    "metadata": {
208 |     "collapsed": false
209 |    },
210 |    "outputs": [],
211 |    "source": [
212 |     "!runipy predict_result.ipynb"
213 |    ]
214 |   },
215 |   {
216 |    "cell_type": "code",
217 |    "execution_count": null,
218 |    "metadata": {
219 |     "collapsed": false
220 |    },
221 |    "outputs": [],
222 |    "source": [
223 |     "!date"
224 |    ]
225 |   }
226 |  ],
227 |  "metadata": {
228 |   "anaconda-cloud": {},
229 |   "hide_input": false,
230 |   "kernelspec": {
231 |    "display_name": "Python 3",
232 |    "language": "python",
233 |    "name": "python3"
234 |   },
235 |   "language_info": {
236 |    "codemirror_mode": {
237 |     "name": "ipython",
238 |     "version": 3
239 |    },
240 |    "file_extension": ".py",
241 |    "mimetype": "text/x-python",
242 |    "name": "python",
243 |    "nbconvert_exporter": "python",
244 |    "pygments_lexer": "ipython3",
245 |    "version": "3.5.1"
246 |   }
247 |  },
248 |  "nbformat": 4,
249 |  "nbformat_minor": 2
250 | }
251 | 


--------------------------------------------------------------------------------
/code/split_samples.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 123,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "from os.path import exists, join\n",
 12 |     "from os import mkdir\n",
 13 |     "from datetime import datetime, timedelta, date\n",
 14 |     "from collections import namedtuple\n",
 15 |     "\n",
 16 |     "import numpy as np\n",
 17 |     "import pandas as pd\n",
 18 |     "import scipy as sp"
 19 |    ]
 20 |   },
 21 |   {
 22 |    "cell_type": "code",
 23 |    "execution_count": 124,
 24 |    "metadata": {
 25 |     "collapsed": true
 26 |    },
 27 |    "outputs": [],
 28 |    "source": [
 29 |     "train_base_dir = '../dataset/fetures/'"
 30 |    ]
 31 |   },
 32 |   {
 33 |    "cell_type": "markdown",
 34 |    "metadata": {},
 35 |    "source": [
 36 |     "# 移动滑窗划分成9个predict样本 "
 37 |    ]
 38 |   },
 39 |   {
 40 |    "cell_type": "code",
 41 |    "execution_count": 125,
 42 |    "metadata": {
 43 |     "collapsed": false
 44 |    },
 45 |    "outputs": [],
 46 |    "source": [
 47 |     "df = pd.read_csv('../dataset/train.csv', parse_dates=['record_date'], index_col=['record_date'])"
 48 |    ]
 49 |   },
 50 |   {
 51 |    "cell_type": "code",
 52 |    "execution_count": 126,
 53 |    "metadata": {
 54 |     "collapsed": false
 55 |    },
 56 |    "outputs": [],
 57 |    "source": [
 58 |     "begin = df.index.max().date() + timedelta(1)"
 59 |    ]
 60 |   },
 61 |   {
 62 |    "cell_type": "markdown",
 63 |    "metadata": {},
 64 |    "source": [
 65 |     "## 训练样本划分"
 66 |    ]
 67 |   },
 68 |   {
 69 |    "cell_type": "code",
 70 |    "execution_count": 127,
 71 |    "metadata": {
 72 |     "collapsed": true
 73 |    },
 74 |    "outputs": [],
 75 |    "source": [
 76 |     "TimeSpan = namedtuple('TimeSpan', ['train_start', 'train_end', 'predict_start', 'predict_end'])"
 77 |    ]
 78 |   },
 79 |   {
 80 |    "cell_type": "code",
 81 |    "execution_count": 128,
 82 |    "metadata": {
 83 |     "collapsed": true
 84 |    },
 85 |    "outputs": [],
 86 |    "source": [
 87 |     "timespans = []"
 88 |    ]
 89 |   },
 90 |   {
 91 |    "cell_type": "code",
 92 |    "execution_count": 129,
 93 |    "metadata": {
 94 |     "collapsed": true
 95 |    },
 96 |    "outputs": [],
 97 |    "source": [
 98 |     "end = begin - timedelta(1)\n",
 99 |     "for i in range(9):\n",
100 |     "    p_end = date(begin.year, begin.month-i, begin.day)\n",
101 |     "    predict_end = p_end - timedelta(1)\n",
102 |     "    predict_start = date(begin.year, begin.month-i-1, begin.day)\n",
103 |     "    train_end = predict_start - timedelta(1)\n",
104 |     "    train_start = date(predict_start.year-1, predict_start.month, predict_start.day)\n",
105 |     "    span = TimeSpan(train_start, train_end, predict_start, predict_end)\n",
106 |     "#     print(span)\n",
107 |     "    timespans.append(span)"
108 |    ]
109 |   },
110 |   {
111 |    "cell_type": "markdown",
112 |    "metadata": {},
113 |    "source": [
114 |     "##### 创建文件架 "
115 |    ]
116 |   },
117 |   {
118 |    "cell_type": "code",
119 |    "execution_count": 130,
120 |    "metadata": {
121 |     "collapsed": false
122 |    },
123 |    "outputs": [],
124 |    "source": [
125 |     "for span in timespans:\n",
126 |     "    dir_n = str(span.predict_start.month)\n",
127 |     "    path = join(train_base_dir, dir_n)\n",
128 |     "    if not exists(path):\n",
129 |     "        mkdir(path)"
130 |    ]
131 |   },
132 |   {
133 |    "cell_type": "code",
134 |    "execution_count": 131,
135 |    "metadata": {
136 |     "collapsed": false
137 |    },
138 |    "outputs": [],
139 |    "source": [
140 |     "span = timespans[0]"
141 |    ]
142 |   },
143 |   {
144 |    "cell_type": "code",
145 |    "execution_count": 132,
146 |    "metadata": {
147 |     "collapsed": true
148 |    },
149 |    "outputs": [],
150 |    "source": [
151 |     "def create_train_sample(timespan):  \n",
152 |     "    train = df.loc[str(span.train_start):str(span.train_end)]    \n",
153 |     "    train_predict = df.loc[str(span.predict_start):str(span.predict_end)].reset_index()\n",
154 |     "    train_predict = train_predict.groupby('record_date')[['power_consumption']].sum()\n",
155 |     "    train_predict['predict_power_consumption'] = train_predict.power_consumption.astype(np.int64)\n",
156 |     "    train_predict.drop('power_consumption', axis=1, inplace=True)\n",
157 |     "    train_predict.index.name = 'predict_date'\n",
158 |     "    train_path = join(train_base_dir, str(span.predict_start.month), 'train.csv')\n",
159 |     "    train_predict_path = join(train_base_dir, str(span.predict_start.month), 'train_predict.csv')\n",
160 |     "    train.to_csv(train_path)\n",
161 |     "    train.info()\n",
162 |     "    train_predict.to_csv(train_predict_path)\n",
163 |     "    train_predict.info()"
164 |    ]
165 |   },
166 |   {
167 |    "cell_type": "code",
168 |    "execution_count": 133,
169 |    "metadata": {
170 |     "collapsed": false
171 |    },
172 |    "outputs": [
173 |     {
174 |      "name": "stdout",
175 |      "output_type": "stream",
176 |      "text": [
177 |       "<class 'pandas.core.frame.DataFrame'>\n",
178 |       "DatetimeIndex: 532156 entries, 2015-09-01 to 2016-08-31\n",
179 |       "Data columns (total 2 columns):\n",
180 |       "user_id              532156 non-null int64\n",
181 |       "power_consumption    532156 non-null int64\n",
182 |       "dtypes: int64(2)\n",
183 |       "memory usage: 12.2 MB\n",
184 |       "<class 'pandas.core.frame.DataFrame'>\n",
185 |       "DatetimeIndex: 30 entries, 2016-09-01 to 2016-09-30\n",
186 |       "Data columns (total 1 columns):\n",
187 |       "predict_power_consumption    30 non-null int64\n",
188 |       "dtypes: int64(1)\n",
189 |       "memory usage: 480.0 bytes\n",
190 |       "<class 'pandas.core.frame.DataFrame'>\n",
191 |       "DatetimeIndex: 532156 entries, 2015-08-01 to 2016-07-31\n",
192 |       "Data columns (total 2 columns):\n",
193 |       "user_id              532156 non-null int64\n",
194 |       "power_consumption    532156 non-null int64\n",
195 |       "dtypes: int64(2)\n",
196 |       "memory usage: 12.2 MB\n",
197 |       "<class 'pandas.core.frame.DataFrame'>\n",
198 |       "DatetimeIndex: 31 entries, 2016-08-01 to 2016-08-31\n",
199 |       "Data columns (total 1 columns):\n",
200 |       "predict_power_consumption    31 non-null int64\n",
201 |       "dtypes: int64(1)\n",
202 |       "memory usage: 496.0 bytes\n",
203 |       "<class 'pandas.core.frame.DataFrame'>\n",
204 |       "DatetimeIndex: 532153 entries, 2015-07-01 to 2016-06-30\n",
205 |       "Data columns (total 2 columns):\n",
206 |       "user_id              532153 non-null int64\n",
207 |       "power_consumption    532153 non-null int64\n",
208 |       "dtypes: int64(2)\n",
209 |       "memory usage: 12.2 MB\n",
210 |       "<class 'pandas.core.frame.DataFrame'>\n",
211 |       "DatetimeIndex: 31 entries, 2016-07-01 to 2016-07-31\n",
212 |       "Data columns (total 1 columns):\n",
213 |       "predict_power_consumption    31 non-null int64\n",
214 |       "dtypes: int64(1)\n",
215 |       "memory usage: 496.0 bytes\n",
216 |       "<class 'pandas.core.frame.DataFrame'>\n",
217 |       "DatetimeIndex: 532150 entries, 2015-06-01 to 2016-05-31\n",
218 |       "Data columns (total 2 columns):\n",
219 |       "user_id              532150 non-null int64\n",
220 |       "power_consumption    532150 non-null int64\n",
221 |       "dtypes: int64(2)\n",
222 |       "memory usage: 12.2 MB\n",
223 |       "<class 'pandas.core.frame.DataFrame'>\n",
224 |       "DatetimeIndex: 30 entries, 2016-06-01 to 2016-06-30\n",
225 |       "Data columns (total 1 columns):\n",
226 |       "predict_power_consumption    30 non-null int64\n",
227 |       "dtypes: int64(1)\n",
228 |       "memory usage: 480.0 bytes\n",
229 |       "<class 'pandas.core.frame.DataFrame'>\n",
230 |       "DatetimeIndex: 532150 entries, 2015-05-01 to 2016-04-30\n",
231 |       "Data columns (total 2 columns):\n",
232 |       "user_id              532150 non-null int64\n",
233 |       "power_consumption    532150 non-null int64\n",
234 |       "dtypes: int64(2)\n",
235 |       "memory usage: 12.2 MB\n",
236 |       "<class 'pandas.core.frame.DataFrame'>\n",
237 |       "DatetimeIndex: 31 entries, 2016-05-01 to 2016-05-31\n",
238 |       "Data columns (total 1 columns):\n",
239 |       "predict_power_consumption    31 non-null int64\n",
240 |       "dtypes: int64(1)\n",
241 |       "memory usage: 496.0 bytes\n",
242 |       "<class 'pandas.core.frame.DataFrame'>\n",
243 |       "DatetimeIndex: 532148 entries, 2015-04-01 to 2016-03-31\n",
244 |       "Data columns (total 2 columns):\n",
245 |       "user_id              532148 non-null int64\n",
246 |       "power_consumption    532148 non-null int64\n",
247 |       "dtypes: int64(2)\n",
248 |       "memory usage: 12.2 MB\n",
249 |       "<class 'pandas.core.frame.DataFrame'>\n",
250 |       "DatetimeIndex: 30 entries, 2016-04-01 to 2016-04-30\n",
251 |       "Data columns (total 1 columns):\n",
252 |       "predict_power_consumption    30 non-null int64\n",
253 |       "dtypes: int64(1)\n",
254 |       "memory usage: 480.0 bytes\n",
255 |       "<class 'pandas.core.frame.DataFrame'>\n",
256 |       "DatetimeIndex: 532147 entries, 2015-03-01 to 2016-02-29\n",
257 |       "Data columns (total 2 columns):\n",
258 |       "user_id              532147 non-null int64\n",
259 |       "power_consumption    532147 non-null int64\n",
260 |       "dtypes: int64(2)\n",
261 |       "memory usage: 12.2 MB\n",
262 |       "<class 'pandas.core.frame.DataFrame'>\n",
263 |       "DatetimeIndex: 31 entries, 2016-03-01 to 2016-03-31\n",
264 |       "Data columns (total 1 columns):\n",
265 |       "predict_power_consumption    31 non-null int64\n",
266 |       "dtypes: int64(1)\n",
267 |       "memory usage: 496.0 bytes\n",
268 |       "<class 'pandas.core.frame.DataFrame'>\n",
269 |       "DatetimeIndex: 530692 entries, 2015-02-01 to 2016-01-31\n",
270 |       "Data columns (total 2 columns):\n",
271 |       "user_id              530692 non-null int64\n",
272 |       "power_consumption    530692 non-null int64\n",
273 |       "dtypes: int64(2)\n",
274 |       "memory usage: 12.1 MB\n",
275 |       "<class 'pandas.core.frame.DataFrame'>\n",
276 |       "DatetimeIndex: 29 entries, 2016-02-01 to 2016-02-29\n",
277 |       "Data columns (total 1 columns):\n",
278 |       "predict_power_consumption    29 non-null int64\n",
279 |       "dtypes: int64(1)\n",
280 |       "memory usage: 464.0 bytes\n",
281 |       "<class 'pandas.core.frame.DataFrame'>\n",
282 |       "DatetimeIndex: 530693 entries, 2015-01-01 to 2015-12-31\n",
283 |       "Data columns (total 2 columns):\n",
284 |       "user_id              530693 non-null int64\n",
285 |       "power_consumption    530693 non-null int64\n",
286 |       "dtypes: int64(2)\n",
287 |       "memory usage: 12.1 MB\n",
288 |       "<class 'pandas.core.frame.DataFrame'>\n",
289 |       "DatetimeIndex: 31 entries, 2016-01-01 to 2016-01-31\n",
290 |       "Data columns (total 1 columns):\n",
291 |       "predict_power_consumption    31 non-null int64\n",
292 |       "dtypes: int64(1)\n",
293 |       "memory usage: 496.0 bytes\n"
294 |      ]
295 |     }
296 |    ],
297 |    "source": [
298 |     "for span in timespans:\n",
299 |     "    create_train_sample(span)"
300 |    ]
301 |   },
302 |   {
303 |    "cell_type": "code",
304 |    "execution_count": 138,
305 |    "metadata": {
306 |     "collapsed": true
307 |    },
308 |    "outputs": [],
309 |    "source": [
310 |     "test = df.loc['2015/10/1':'2016/9/30']"
311 |    ]
312 |   },
313 |   {
314 |    "cell_type": "code",
315 |    "execution_count": 139,
316 |    "metadata": {
317 |     "collapsed": false
318 |    },
319 |    "outputs": [],
320 |    "source": [
321 |     "index = pd.date_range('2016/10/1', '2016/10/31')\n",
322 |     "test_predict = pd.DataFrame(0, index=index, columns=['predict_power_consumption'])\n",
323 |     "test_predict.index.name = 'predict_date'"
324 |    ]
325 |   },
326 |   {
327 |    "cell_type": "code",
328 |    "execution_count": 140,
329 |    "metadata": {
330 |     "collapsed": false
331 |    },
332 |    "outputs": [
333 |     {
334 |      "name": "stdout",
335 |      "output_type": "stream",
336 |      "text": [
337 |       "<class 'pandas.core.frame.DataFrame'>\n",
338 |       "DatetimeIndex: 532159 entries, 2015-10-01 to 2016-09-28\n",
339 |       "Data columns (total 2 columns):\n",
340 |       "user_id              532159 non-null int64\n",
341 |       "power_consumption    532159 non-null int64\n",
342 |       "dtypes: int64(2)\n",
343 |       "memory usage: 12.2 MB\n"
344 |      ]
345 |     }
346 |    ],
347 |    "source": [
348 |     "test.to_csv('../dataset/fetures/test.csv')\n",
349 |     "test.info()"
350 |    ]
351 |   },
352 |   {
353 |    "cell_type": "code",
354 |    "execution_count": 142,
355 |    "metadata": {
356 |     "collapsed": false
357 |    },
358 |    "outputs": [
359 |     {
360 |      "name": "stdout",
361 |      "output_type": "stream",
362 |      "text": [
363 |       "<class 'pandas.core.frame.DataFrame'>\n",
364 |       "DatetimeIndex: 31 entries, 2016-10-01 to 2016-10-31\n",
365 |       "Freq: D\n",
366 |       "Data columns (total 1 columns):\n",
367 |       "predict_power_consumption    31 non-null int64\n",
368 |       "dtypes: int64(1)\n",
369 |       "memory usage: 496.0 bytes\n"
370 |      ]
371 |     }
372 |    ],
373 |    "source": [
374 |     "test_predict.to_csv('../dataset/fetures/test_predict.csv')\n",
375 |     "test_predict.info()"
376 |    ]
377 |   }
378 |  ],
379 |  "metadata": {
380 |   "anaconda-cloud": {},
381 |   "hide_input": false,
382 |   "kernelspec": {
383 |    "display_name": "Python 3",
384 |    "language": "python",
385 |    "name": "python3"
386 |   },
387 |   "language_info": {
388 |    "codemirror_mode": {
389 |     "name": "ipython",
390 |     "version": 3
391 |    },
392 |    "file_extension": ".py",
393 |    "mimetype": "text/x-python",
394 |    "name": "python",
395 |    "nbconvert_exporter": "python",
396 |    "pygments_lexer": "ipython3",
397 |    "version": "3.5.1"
398 |   }
399 |  },
400 |  "nbformat": 4,
401 |  "nbformat_minor": 2
402 | }
403 | 


--------------------------------------------------------------------------------
/code/train_model.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 35,
  6 |    "metadata": {
  7 |     "collapsed": false
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import pickle\n",
 12 |     "from datetime import datetime\n",
 13 |     "\n",
 14 |     "import numpy as np\n",
 15 |     "import scipy as sp\n",
 16 |     "import pandas as pd\n",
 17 |     "\n",
 18 |     "from sklearn.ensemble import RandomForestRegressor\n",
 19 |     "from sklearn.metrics import r2_score, roc_auc_score\n",
 20 |     "from sklearn import cross_validation\n",
 21 |     "from sklearn.model_selection import GridSearchCV\n",
 22 |     "from sklearn.svm import SVR\n",
 23 |     "import mlxtend"
 24 |    ]
 25 |   },
 26 |   {
 27 |    "cell_type": "code",
 28 |    "execution_count": 36,
 29 |    "metadata": {
 30 |     "collapsed": true
 31 |    },
 32 |    "outputs": [],
 33 |    "source": [
 34 |     "from mlxtend.regressor import StackingRegressor\n",
 35 |     "from mlxtend.data import boston_housing_data\n",
 36 |     "from sklearn.linear_model import LinearRegression\n",
 37 |     "from sklearn.linear_model import Ridge\n",
 38 |     "from sklearn.svm import SVR\n",
 39 |     "from sklearn.ensemble import ExtraTreesRegressor"
 40 |    ]
 41 |   },
 42 |   {
 43 |    "cell_type": "code",
 44 |    "execution_count": 37,
 45 |    "metadata": {
 46 |     "collapsed": true
 47 |    },
 48 |    "outputs": [],
 49 |    "source": [
 50 |     "def loss_score(predict, real):\n",
 51 |     "    f = (real - predict)/real\n",
 52 |     "    n = len(f)\n",
 53 |     "    f = f.replace([np.nan, -np.nan], 0)\n",
 54 |     "    score = 1 - np.abs(f).sum()/n\n",
 55 |     "    return score "
 56 |    ]
 57 |   },
 58 |   {
 59 |    "cell_type": "code",
 60 |    "execution_count": 38,
 61 |    "metadata": {
 62 |     "collapsed": true
 63 |    },
 64 |    "outputs": [],
 65 |    "source": [
 66 |     "model_path = '../dataset/fetures/model.pkl'"
 67 |    ]
 68 |   },
 69 |   {
 70 |    "cell_type": "code",
 71 |    "execution_count": 39,
 72 |    "metadata": {
 73 |     "collapsed": true
 74 |    },
 75 |    "outputs": [],
 76 |    "source": [
 77 |     "fetures = pd.read_csv('../dataset/fetures/train_feture.csv')"
 78 |    ]
 79 |   },
 80 |   {
 81 |    "cell_type": "code",
 82 |    "execution_count": 40,
 83 |    "metadata": {
 84 |     "collapsed": false
 85 |    },
 86 |    "outputs": [],
 87 |    "source": [
 88 |     "train_x = fetures.drop(['predict_date', 'predict_power_consumption'], axis=1)"
 89 |    ]
 90 |   },
 91 |   {
 92 |    "cell_type": "code",
 93 |    "execution_count": 41,
 94 |    "metadata": {
 95 |     "collapsed": false
 96 |    },
 97 |    "outputs": [],
 98 |    "source": [
 99 |     "train_y = fetures['predict_power_consumption']"
100 |    ]
101 |   },
102 |   {
103 |    "cell_type": "markdown",
104 |    "metadata": {},
105 |    "source": [
106 |     "###  使用 gridsearch 调参"
107 |    ]
108 |   },
109 |   {
110 |    "cell_type": "code",
111 |    "execution_count": 43,
112 |    "metadata": {
113 |     "collapsed": false
114 |    },
115 |    "outputs": [],
116 |    "source": [
117 |     "parameters = {\n",
118 |     "    'n_estimators': [10, 100, 200, 300, 500],\n",
119 |     "#     'criterion': ['mse', 'mae'],\n",
120 |     "    'max_features': [10, 50, 100, 200, 'sqrt']\n",
121 |     "}"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "code",
126 |    "execution_count": 44,
127 |    "metadata": {
128 |     "collapsed": true
129 |    },
130 |    "outputs": [],
131 |    "source": [
132 |     "# model = RandomForestRegressor(n_jobs=-1, oob_score=True)"
133 |    ]
134 |   },
135 |   {
136 |    "cell_type": "code",
137 |    "execution_count": 45,
138 |    "metadata": {
139 |     "collapsed": false
140 |    },
141 |    "outputs": [],
142 |    "source": [
143 |     "# cv = cross_validation.ShuffleSplit(train_x.shape[0], n_iter=8, test_size=0.1, random_state=0)\n",
144 |     "# clf = GridSearchCV(model, parameters, cv=cv)"
145 |    ]
146 |   },
147 |   {
148 |    "cell_type": "code",
149 |    "execution_count": 46,
150 |    "metadata": {
151 |     "collapsed": true
152 |    },
153 |    "outputs": [],
154 |    "source": [
155 |     "# clf.fit(train_x.fillna(0), train_y)"
156 |    ]
157 |   },
158 |   {
159 |    "cell_type": "code",
160 |    "execution_count": 47,
161 |    "metadata": {
162 |     "collapsed": false
163 |    },
164 |    "outputs": [],
165 |    "source": [
166 |     "# model = RandomForestRegressor(n_jobs=-1, oob_score=True, random_state=666, **clf.best_params_)"
167 |    ]
168 |   },
169 |   {
170 |    "cell_type": "code",
171 |    "execution_count": 48,
172 |    "metadata": {
173 |     "collapsed": false
174 |    },
175 |    "outputs": [],
176 |    "source": [
177 |     "# cv = cross_validation.ShuffleSplit(train_x.shape[0], n_iter=5, test_size=0.1, random_state=666)\n",
178 |     "# r2_score = cross_validation.cross_val_score(model, train_x.fillna(0), train_y, cv=cv)\n",
179 |     "# r2_score"
180 |    ]
181 |   },
182 |   {
183 |    "cell_type": "code",
184 |    "execution_count": 288,
185 |    "metadata": {
186 |     "collapsed": true
187 |    },
188 |    "outputs": [],
189 |    "source": [
190 |     "cv = cross_validation.ShuffleSplit(train_x.shape[0], n_iter=10, test_size=0.2, random_state=666)"
191 |    ]
192 |   },
193 |   {
194 |    "cell_type": "code",
195 |    "execution_count": 290,
196 |    "metadata": {
197 |     "collapsed": true
198 |    },
199 |    "outputs": [],
200 |    "source": [
201 |     "forest = RandomForestRegressor(\n",
202 |     "    n_jobs=-1,\n",
203 |     "#     criterion='mae',\n",
204 |     "    random_state=1,\n",
205 |     "    oob_score=True,\n",
206 |     "    n_estimators=100,\n",
207 |     "    max_features='auto')\n",
208 |     "fores = RandomForestRegressor(\n",
209 |     "    n_jobs=-1,\n",
210 |     "    random_state=2,\n",
211 |     "    oob_score=True,\n",
212 |     "    n_estimators=300,\n",
213 |     "    max_features=300)\n",
214 |     "extract = ExtraTreesRegressor(\n",
215 |     "    n_jobs=-1,\n",
216 |     "    random_state=3,\n",
217 |     "#     criterion='mae',\n",
218 |     "    bootstrap=True,\n",
219 |     "    oob_score=True,\n",
220 |     "    n_estimators=100,\n",
221 |     "    max_features='auto')\n",
222 |     "extrac = ExtraTreesRegressor(\n",
223 |     "    n_jobs=-1,\n",
224 |     "    random_state=4,\n",
225 |     "    bootstrap=True,\n",
226 |     "    oob_score=True,\n",
227 |     "    n_estimators=300,\n",
228 |     "    max_features=300)"
229 |    ]
230 |   },
231 |   {
232 |    "cell_type": "code",
233 |    "execution_count": 292,
234 |    "metadata": {
235 |     "collapsed": true
236 |    },
237 |    "outputs": [],
238 |    "source": [
239 |     "clfs = [forest, fores, extract, extrac]"
240 |    ]
241 |   },
242 |   {
243 |    "cell_type": "code",
244 |    "execution_count": 293,
245 |    "metadata": {
246 |     "collapsed": true
247 |    },
248 |    "outputs": [],
249 |    "source": [
250 |     "params = {\n",
251 |     "    'forest__n-estimators': [10, 50, 100],\n",
252 |     "    'forest__max-features': [10, 50, 100],\n",
253 |     "    'fores__n-estimators': [100, 200, 500],\n",
254 |     "    'fores__max-features': [100, 200, 300],\n",
255 |     "    'extract__n-estimators': [10, 50, 100],\n",
256 |     "    'extract__max-features': [10, 50, 100],\n",
257 |     "    'extrac__n-estimators': [100, 200, 500],\n",
258 |     "    'extrac__max-features': [100, 200, 300],\n",
259 |     "}"
260 |    ]
261 |   },
262 |   {
263 |    "cell_type": "code",
264 |    "execution_count": 294,
265 |    "metadata": {
266 |     "collapsed": false
267 |    },
268 |    "outputs": [],
269 |    "source": [
270 |     "meta = LinearRegression(n_jobs=-1)"
271 |    ]
272 |   },
273 |   {
274 |    "cell_type": "code",
275 |    "execution_count": 295,
276 |    "metadata": {
277 |     "collapsed": false
278 |    },
279 |    "outputs": [],
280 |    "source": [
281 |     "stacker = StackingRegressor(regressors=clfs, meta_regressor=meta, verbose=True)"
282 |    ]
283 |   },
284 |   {
285 |    "cell_type": "code",
286 |    "execution_count": 296,
287 |    "metadata": {
288 |     "collapsed": false
289 |    },
290 |    "outputs": [],
291 |    "source": [
292 |     "grid = GridSearchCV(estimator=stacker, param_grid=params, cv=cv)"
293 |    ]
294 |   },
295 |   {
296 |    "cell_type": "code",
297 |    "execution_count": 297,
298 |    "metadata": {
299 |     "collapsed": false
300 |    },
301 |    "outputs": [
302 |     {
303 |      "data": {
304 |       "text/plain": [
305 |        "{'bootstrap': True,\n",
306 |        " 'criterion': 'mse',\n",
307 |        " 'max_depth': None,\n",
308 |        " 'max_features': 'auto',\n",
309 |        " 'max_leaf_nodes': None,\n",
310 |        " 'min_impurity_split': 1e-07,\n",
311 |        " 'min_samples_leaf': 1,\n",
312 |        " 'min_samples_split': 2,\n",
313 |        " 'min_weight_fraction_leaf': 0.0,\n",
314 |        " 'n_estimators': 100,\n",
315 |        " 'n_jobs': -1,\n",
316 |        " 'oob_score': True,\n",
317 |        " 'random_state': 3,\n",
318 |        " 'verbose': 0,\n",
319 |        " 'warm_start': False}"
320 |       ]
321 |      },
322 |      "execution_count": 297,
323 |      "metadata": {},
324 |      "output_type": "execute_result"
325 |     }
326 |    ],
327 |    "source": [
328 |     "extract.get_params()"
329 |    ]
330 |   },
331 |   {
332 |    "cell_type": "code",
333 |    "execution_count": 298,
334 |    "metadata": {
335 |     "collapsed": true
336 |    },
337 |    "outputs": [],
338 |    "source": [
339 |     "# grid.fit(train_x.fillna(0), train_y)"
340 |    ]
341 |   },
342 |   {
343 |    "cell_type": "code",
344 |    "execution_count": 300,
345 |    "metadata": {
346 |     "collapsed": true
347 |    },
348 |    "outputs": [
349 |     {
350 |      "name": "stdout",
351 |      "output_type": "stream",
352 |      "text": [
353 |       "Fitting 4 regressors...\n",
354 |       "Fitting regressor1: randomforestregressor (1/4)\n",
355 |       "Fitting 4 regressors...\n",
356 |       "Fitting regressor1: randomforestregressor (1/4)\n",
357 |       "Fitting 4 regressors...\n",
358 |       "Fitting regressor1: randomforestregressor (1/4)\n",
359 |       "Fitting 4 regressors...\n",
360 |       "Fitting regressor1: randomforestregressor (1/4)\n",
361 |       "Fitting regressor2: randomforestregressor (2/4)\n",
362 |       "Fitting regressor2: randomforestregressor (2/4)\n",
363 |       "Fitting regressor2: randomforestregressor (2/4)\n",
364 |       "Fitting regressor2: randomforestregressor (2/4)\n",
365 |       "Fitting regressor3: extratreesregressor (3/4)\n",
366 |       "Fitting regressor3: extratreesregressor (3/4)\n",
367 |       "Fitting regressor3: extratreesregressor (3/4)\n",
368 |       "Fitting regressor3: extratreesregressor (3/4)\n",
369 |       "Fitting regressor4: extratreesregressor (4/4)\n",
370 |       "Fitting regressor4: extratreesregressor (4/4)\n",
371 |       "Fitting regressor4: extratreesregressor (4/4)\n",
372 |       "Fitting regressor4: extratreesregressor (4/4)\n",
373 |       "Fitting 4 regressors...\n",
374 |       "Fitting regressor1: randomforestregressor (1/4)\n",
375 |       "Fitting 4 regressors...\n",
376 |       "Fitting regressor1: randomforestregressor (1/4)\n",
377 |       "Fitting 4 regressors...\n",
378 |       "Fitting regressor1: randomforestregressor (1/4)\n",
379 |       "Fitting 4 regressors...\n",
380 |       "Fitting regressor1: randomforestregressor (1/4)\n",
381 |       "Fitting regressor2: randomforestregressor (2/4)\n",
382 |       "Fitting regressor2: randomforestregressor (2/4)\n",
383 |       "Fitting regressor2: randomforestregressor (2/4)\n",
384 |       "Fitting regressor2: randomforestregressor (2/4)\n",
385 |       "Fitting regressor3: extratreesregressor (3/4)\n",
386 |       "Fitting regressor3: extratreesregressor (3/4)\n",
387 |       "Fitting regressor3: extratreesregressor (3/4)\n",
388 |       "Fitting regressor3: extratreesregressor (3/4)\n",
389 |       "Fitting regressor4: extratreesregressor (4/4)\n",
390 |       "Fitting regressor4: extratreesregressor (4/4)\n",
391 |       "Fitting regressor4: extratreesregressor (4/4)\n",
392 |       "Fitting regressor4: extratreesregressor (4/4)\n",
393 |       "Fitting 4 regressors...\n",
394 |       "Fitting regressor1: randomforestregressor (1/4)\n",
395 |       "Fitting 4 regressors...\n",
396 |       "Fitting regressor1: randomforestregressor (1/4)\n",
397 |       "Fitting regressor2: randomforestregressor (2/4)\n",
398 |       "Fitting regressor2: randomforestregressor (2/4)\n",
399 |       "Fitting regressor3: extratreesregressor (3/4)\n",
400 |       "Fitting regressor3: extratreesregressor (3/4)\n",
401 |       "Fitting regressor4: extratreesregressor (4/4)\n",
402 |       "Fitting regressor4: extratreesregressor (4/4)\n"
403 |      ]
404 |     },
405 |     {
406 |      "name": "stderr",
407 |      "output_type": "stream",
408 |      "text": [
409 |       "[Parallel(n_jobs=-1)]: Done  10 out of  10 | elapsed:  2.1min finished\n"
410 |      ]
411 |     }
412 |    ],
413 |    "source": [
414 |     "r2_score = cross_validation.cross_val_score(stacker, train_x, train_y, cv=cv, verbose=True, n_jobs=-1)"
415 |    ]
416 |   },
417 |   {
418 |    "cell_type": "code",
419 |    "execution_count": 301,
420 |    "metadata": {
421 |     "collapsed": true
422 |    },
423 |    "outputs": [],
424 |    "source": [
425 |     "# grid.fit(train_x.fillna(0), train_y)"
426 |    ]
427 |   },
428 |   {
429 |    "cell_type": "code",
430 |    "execution_count": 302,
431 |    "metadata": {
432 |     "collapsed": false
433 |    },
434 |    "outputs": [
435 |     {
436 |      "data": {
437 |       "text/plain": [
438 |        "(0.78898896165565757, 0.86008654041103993, 0.68182410751042843)"
439 |       ]
440 |      },
441 |      "execution_count": 302,
442 |      "metadata": {},
443 |      "output_type": "execute_result"
444 |     }
445 |    ],
446 |    "source": [
447 |     "r2_score.mean(), r2_score.max(), r2_score.min()"
448 |    ]
449 |   },
450 |   {
451 |    "cell_type": "code",
452 |    "execution_count": 303,
453 |    "metadata": {
454 |     "collapsed": false
455 |    },
456 |    "outputs": [
457 |     {
458 |      "name": "stdout",
459 |      "output_type": "stream",
460 |      "text": [
461 |       "Fitting 4 regressors...\n",
462 |       "Fitting regressor1: randomforestregressor (1/4)\n",
463 |       "Fitting regressor2: randomforestregressor (2/4)\n",
464 |       "Fitting regressor3: extratreesregressor (3/4)\n",
465 |       "Fitting regressor4: extratreesregressor (4/4)\n"
466 |      ]
467 |     },
468 |     {
469 |      "data": {
470 |       "text/plain": [
471 |        "StackingRegressor(meta_regressor=LinearRegression(copy_X=True, fit_intercept=True, n_jobs=-1, normalize=False),\n",
472 |        "         regressors=[RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
473 |        "           max_features='auto', max_leaf_nodes=None,\n",
474 |        "           min_impurity_split=1e-07, min_samples_leaf=1,\n",
475 |        "           min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
476 |        "           n_estimators=100, n_jobs=-1, oob_s...n_estimators=300, n_jobs=-1,\n",
477 |        "          oob_score=True, random_state=4, verbose=0, warm_start=False)],\n",
478 |        "         verbose=True)"
479 |       ]
480 |      },
481 |      "execution_count": 303,
482 |      "metadata": {},
483 |      "output_type": "execute_result"
484 |     }
485 |    ],
486 |    "source": [
487 |     "stacker.fit(train_x.fillna(0), train_y)"
488 |    ]
489 |   },
490 |   {
491 |    "cell_type": "code",
492 |    "execution_count": 304,
493 |    "metadata": {
494 |     "collapsed": false
495 |    },
496 |    "outputs": [],
497 |    "source": [
498 |     "r2_all_scorr = stacker.score(train_x.fillna(0), train_y)"
499 |    ]
500 |   },
501 |   {
502 |    "cell_type": "code",
503 |    "execution_count": null,
504 |    "metadata": {
505 |     "collapsed": true
506 |    },
507 |    "outputs": [],
508 |    "source": [
509 |     "# importance_df = pd.DataFrame(model.feature_importances_, index=train_x.columns)\n",
510 |     "\n",
511 |     "# importance_df.sort_values(0, ascending=False, inplace=True)\n",
512 |     "\n",
513 |     "# importance_top20 = ' , '.join(['{}:{} '.format(x, importance_df.loc[x].values[0]) for x in importance_df.index][:20])\n",
514 |     "# importance_top20"
515 |    ]
516 |   },
517 |   {
518 |    "cell_type": "code",
519 |    "execution_count": 305,
520 |    "metadata": {
521 |     "collapsed": false
522 |    },
523 |    "outputs": [
524 |     {
525 |      "data": {
526 |       "text/plain": [
527 |        "0.9876500688451734"
528 |       ]
529 |      },
530 |      "execution_count": 305,
531 |      "metadata": {},
532 |      "output_type": "execute_result"
533 |     }
534 |    ],
535 |    "source": [
536 |     "predict_result = stacker.predict(train_x.fillna(0))\n",
537 |     "loss_s  = loss_score(predict_result, train_y)\n",
538 |     "loss_s"
539 |    ]
540 |   },
541 |   {
542 |    "cell_type": "code",
543 |    "execution_count": 306,
544 |    "metadata": {
545 |     "collapsed": false
546 |    },
547 |    "outputs": [
548 |     {
549 |      "data": {
550 |       "text/plain": [
551 |        "(0.9876500688451734,\n",
552 |        " 0.98078256144556675,\n",
553 |        " 0.78898896165565757,\n",
554 |        " 0.86008654041103993,\n",
555 |        " 0.68182410751042843)"
556 |       ]
557 |      },
558 |      "execution_count": 306,
559 |      "metadata": {},
560 |      "output_type": "execute_result"
561 |     }
562 |    ],
563 |    "source": [
564 |     "result = loss_s, r2_all_scorr, r2_score.mean(), r2_score.max(), r2_score.min()\n",
565 |     "result"
566 |    ]
567 |   },
568 |   {
569 |    "cell_type": "code",
570 |    "execution_count": 307,
571 |    "metadata": {
572 |     "collapsed": false
573 |    },
574 |    "outputs": [
575 |     {
576 |      "name": "stdout",
577 |      "output_type": "stream",
578 |      "text": [
579 |       "2017-06-13 07:55:11.228802 result: (0.9876500688451734, 0.98078256144556675, 0.78898896165565757, 0.86008654041103993, 0.68182410751042843)\n"
580 |      ]
581 |     }
582 |    ],
583 |    "source": [
584 |     "print(datetime.now(), 'result:', result)\n",
585 |     "print(\n",
586 |     "    datetime.now(),\n",
587 |     "    'result:', result,\n",
588 |     "    file=open('result.txt', 'a+'),\n",
589 |     "    sep=' ',\n",
590 |     "    end='\\n')"
591 |    ]
592 |   },
593 |   {
594 |    "cell_type": "code",
595 |    "execution_count": 308,
596 |    "metadata": {
597 |     "collapsed": false
598 |    },
599 |    "outputs": [],
600 |    "source": [
601 |     "pickle.dump(stacker, open(model_path, 'wb'))"
602 |    ]
603 |   }
604 |  ],
605 |  "metadata": {
606 |   "anaconda-cloud": {},
607 |   "hide_input": false,
608 |   "kernelspec": {
609 |    "display_name": "Python 3",
610 |    "language": "python",
611 |    "name": "python3"
612 |   },
613 |   "language_info": {
614 |    "codemirror_mode": {
615 |     "name": "ipython",
616 |     "version": 3
617 |    },
618 |    "file_extension": ".py",
619 |    "mimetype": "text/x-python",
620 |    "name": "python",
621 |    "nbconvert_exporter": "python",
622 |    "pygments_lexer": "ipython3",
623 |    "version": "3.5.1"
624 |   }
625 |  },
626 |  "nbformat": 4,
627 |  "nbformat_minor": 2
628 | }
629 | 


--------------------------------------------------------------------------------
/code/trick_extract.py:
--------------------------------------------------------------------------------
  1 | 
  2 | # coding: utf-8
  3 | 
  4 | # In[59]:
  5 | 
  6 | import sys
  7 | from datetime import datetime, timedelta
  8 | 
  9 | import numpy as np
 10 | import pandas as pd
 11 | import scipy as sp
 12 | 
 13 | 
 14 | # In[60]:
 15 | 
 16 | def loss_score(predict, real):
 17 |     f = (real - predict)/real
 18 |     n = len(f)
 19 |     f = f.replace([np.nan, -np.nan], 0)
 20 |     score = 1 - np.abs(f).sum()/n
 21 |     return score 
 22 | # from sklearn.metrics import r2_score
 23 | # loss_score = r2_score
 24 | 
 25 | 
 26 | # In[61]:
 27 | 
 28 | if sys.argv[1] == 'test':
 29 |     is_train = False
 30 | else:
 31 |     is_train = True
 32 | 
 33 | 
 34 | # In[62]:
 35 | 
 36 | if is_train:
 37 |     train_path = '../dataset/fetures/train.csv'
 38 |     predict_path = '../dataset/fetures/train_predict.csv'
 39 |     feture_path = '../dataset/fetures/train_feture.csv'
 40 | else:
 41 |     train_path = '../dataset/fetures/test.csv'
 42 |     predict_path = '../dataset/fetures/test_predict.csv'
 43 |     feture_path = '../dataset/fetures/test_feture.csv'
 44 | 
 45 | 
 46 | # In[63]:
 47 | 
 48 | train = pd.read_csv(train_path, parse_dates=['record_date'])
 49 | 
 50 | 
 51 | # In[64]:
 52 | 
 53 | predict = pd.read_csv(predict_path, parse_dates=['predict_date'], index_col=['predict_date'])
 54 | 
 55 | 
 56 | # # 移动划窗规则
 57 | 
 58 | # In[65]:
 59 | 
 60 | days = len(predict)
 61 | 
 62 | 
 63 | # In[66]:
 64 | 
 65 | avg_df = train.groupby('record_date')[['power_consumption']].sum()
 66 | 
 67 | 
 68 | # In[67]:
 69 | 
 70 | avg_df.index = avg_df.index + timedelta(days)
 71 | 
 72 | 
 73 | # In[68]:
 74 | 
 75 | avg_df.index.name = 'predict_date'
 76 | 
 77 | 
 78 | # In[69]:
 79 | 
 80 | avg_df['predict_power_consumption'] = avg_df.power_consumption.astype(int)
 81 | 
 82 | 
 83 | # In[70]:
 84 | 
 85 | rule_predict = pd.DataFrame(avg_df.predict_power_consumption, index=predict.index)
 86 | 
 87 | 
 88 | # In[71]:
 89 | 
 90 | if is_train:
 91 |     print('move windows score', loss_score( rule_predict, predict))
 92 | 
 93 | 
 94 | # # 平均值
 95 | 
 96 | # In[73]:
 97 | 
 98 | train.record_date.unique().shape
 99 | 
100 | 
101 | # In[74]:
102 | 
103 | def avg_score(month, year='2016'):
104 |     df = train.set_index('record_date').loc['{}/{}/1'.format(year, month):]
105 |     days = df.index.unique().shape[0]
106 |     mean_consumption = np.int64(df.power_consumption.sum()/days)
107 | 
108 |     predict['avg'] = mean_consumption
109 | 
110 |     if is_train:
111 |         loss = loss_score(predict.avg, predict.predict_power_consumption)
112 |         print('from {}-{} '.format(year, month), 'avg {} score', mean_consumption, loss_score(predict.predict_power_consumption, predict.avg))
113 | 
114 |     
115 | 
116 | 
117 | # In[75]:
118 | 
119 | for year in [2015, 2016]:
120 |     for month in range(1, 13):
121 |         if not(year == 2016 and month > 6):
122 |             avg_score(month, year)
123 | 
124 | 
125 | # # 选择2015-1月到至今的平均值
126 | 
127 | # In[76]:
128 | 
129 | avg_score(6, 2016) if is_train else avg_score(9, 2016)
130 | 
131 | 
132 | # In[78]:
133 | 
134 | if not is_train:
135 |     predict['predict_power_consumption'] = predict.avg
136 |     predict.index = predict.index.map(lambda x:x.strftime('%Y%m%d'))
137 |     predict.index.name = 'predict_date'
138 |     predict.to_csv('Tianchi_power_predict_table.csv', columns=['predict_power_consumption'])
139 | 
140 | 


--------------------------------------------------------------------------------
/code/weather_crawl.py:
--------------------------------------------------------------------------------
 1 | # -*- coding:utf-8 -*-
 2 | """
 3 | @author: zhanglun <zhanglun.me@gmail.com>
 4 | @github:  mrzhangboss
 5 | @date: 2017/06/09
 6 | 
 7 | """
 8 | from io import StringIO
 9 | import requests
10 | from lxml import etree
11 | 
12 | html_parser = etree.HTMLParser()
13 | 
14 | 
15 | def create_start_urls(province, year, month):
16 |     url = 'http://lishi.tianqi.com/{}/{}{:02}.html'.format(province, year, month)
17 |     return url
18 | 
19 | 
20 | def parse_ul(ul):
21 |     data = ul.xpath('.//a/text()')
22 |     weather = ul.xpath('.//li/text()')
23 |     return data + weather
24 | 
25 | 
26 | def parse_month(elements):
27 |     result = []
28 |     xpath = "//div[@class='tqtongji2']/ul"
29 |     for ul in elements.xpath(xpath)[1:]:
30 |         r = parse_ul(ul)
31 |         result.append(r)
32 |     return result
33 | 
34 | 
35 | if __name__ == '__main__':
36 |     province = 'yangzhong'
37 |     filename = '{}.csv'.format(province)
38 |     for year in range(2015, 2017):
39 |         for month in range(1, 13):
40 |             url = create_start_urls(province, year, month)
41 |             res = requests.get(url)
42 |             elements = etree.parse(StringIO(res.text), parser=html_parser)
43 |             result = parse_month(elements)
44 |             print(len(result), result)
45 |             with open(filename, 'a') as f:
46 |                 f.write('\n'.join((','.join(x) for x in result)))
47 |                 f.write('\n')


--------------------------------------------------------------------------------
/dataset/holiday.csv:
--------------------------------------------------------------------------------
  1 | date,holiday
  2 | 20150102,2
  3 | 20150224,2
  4 | 20150322,1
  5 | 20150101,2
  6 | 20150329,1
  7 | 20150112,0
  8 | 20150122,0
  9 | 20150124,1
 10 | 20150119,0
 11 | 20150222,2
 12 | 20150221,2
 13 | 20150206,0
 14 | 20150214,1
 15 | 20150131,1
 16 | 20150207,1
 17 | 20150315,1
 18 | 20150108,0
 19 | 20150312,0
 20 | 20150305,0
 21 | 20150111,1
 22 | 20150113,0
 23 | 20150307,1
 24 | 20150103,2
 25 | 20150227,0
 26 | 20150319,0
 27 | 20150406,2
 28 | 20150212,0
 29 | 20150127,0
 30 | 20150218,2
 31 | 20150317,0
 32 | 20150311,0
 33 | 20150318,0
 34 | 20150405,2
 35 | 20150208,1
 36 | 20150302,0
 37 | 20150410,0
 38 | 20150316,0
 39 | 20150301,1
 40 | 20150219,2
 41 | 20150216,0
 42 | 20150106,0
 43 | 20150202,0
 44 | 20150120,0
 45 | 20150201,1
 46 | 20150130,0
 47 | 20150105,0
 48 | 20150320,0
 49 | 20150115,0
 50 | 20150225,0
 51 | 20150321,1
 52 | 20150121,0
 53 | 20150223,2
 54 | 20150331,0
 55 | 20150114,0
 56 | 20150117,1
 57 | 20150403,0
 58 | 20150110,1
 59 | 20150107,0
 60 | 20150330,0
 61 | 20150116,0
 62 | 20150213,0
 63 | 20150215,0
 64 | 20150203,0
 65 | 20150125,1
 66 | 20150404,2
 67 | 20150104,0
 68 | 20150314,1
 69 | 20150220,2
 70 | 20150205,0
 71 | 20150306,0
 72 | 20150327,0
 73 | 20150409,0
 74 | 20150226,0
 75 | 20150129,0
 76 | 20150109,0
 77 | 20150128,0
 78 | 20150401,0
 79 | 20150304,0
 80 | 20150310,0
 81 | 20150210,0
 82 | 20150118,1
 83 | 20150204,0
 84 | 20150308,1
 85 | 20150313,0
 86 | 20150123,0
 87 | 20150217,0
 88 | 20150309,0
 89 | 20150325,0
 90 | 20150402,0
 91 | 20150211,0
 92 | 20150328,1
 93 | 20150408,0
 94 | 20150323,0
 95 | 20150209,0
 96 | 20150303,0
 97 | 20150228,0
 98 | 20150324,0
 99 | 20150407,0
100 | 20150326,0
101 | 20150126,0
102 | 20150504,0
103 | 20150512,0
104 | 20150423,0
105 | 20150519,0
106 | 20150603,0
107 | 20150529,0
108 | 20150412,1
109 | 20150413,0
110 | 20150427,0
111 | 20150717,0
112 | 20150613,1
113 | 20150522,0
114 | 20150629,0
115 | 20150526,0
116 | 20150414,0
117 | 20150527,0
118 | 20150719,1
119 | 20150530,1
120 | 20150502,2
121 | 20150626,0
122 | 20150428,0
123 | 20150521,0
124 | 20150624,0
125 | 20150429,0
126 | 20150528,0
127 | 20150422,0
128 | 20150627,1
129 | 20150517,1
130 | 20150419,1
131 | 20150610,0
132 | 20150614,1
133 | 20150424,0
134 | 20150709,0
135 | 20150710,0
136 | 20150417,0
137 | 20150702,0
138 | 20150617,0
139 | 20150415,0
140 | 20150707,0
141 | 20150509,1
142 | 20150606,1
143 | 20150514,0
144 | 20150703,0
145 | 20150704,1
146 | 20150515,0
147 | 20150621,2
148 | 20150525,0
149 | 20150620,2
150 | 20150604,0
151 | 20150615,0
152 | 20150430,0
153 | 20150616,0
154 | 20150418,1
155 | 20150630,0
156 | 20150507,0
157 | 20150622,2
158 | 20150520,0
159 | 20150609,0
160 | 20150625,0
161 | 20150508,0
162 | 20150524,1
163 | 20150612,0
164 | 20150718,1
165 | 20150426,1
166 | 20150516,1
167 | 20150619,0
168 | 20150420,0
169 | 20150711,1
170 | 20150623,0
171 | 20150510,1
172 | 20150605,0
173 | 20150513,0
174 | 20150712,1
175 | 20150716,0
176 | 20150607,1
177 | 20150506,0
178 | 20150705,1
179 | 20150618,0
180 | 20150503,2
181 | 20150715,0
182 | 20150706,0
183 | 20150611,0
184 | 20150505,0
185 | 20150421,0
186 | 20150628,1
187 | 20150411,1
188 | 20150701,0
189 | 20150608,0
190 | 20150602,0
191 | 20150416,0
192 | 20150714,0
193 | 20150511,0
194 | 20150708,0
195 | 20150501,2
196 | 20150713,0
197 | 20150601,0
198 | 20150425,1
199 | 20150523,1
200 | 20150518,0
201 | 20150531,1
202 | 20151025,1
203 | 20150907,0
204 | 20150913,1
205 | 20151003,2
206 | 20150721,0
207 | 20150929,0
208 | 20150924,0
209 | 20150815,1
210 | 20150911,0
211 | 20150825,0
212 | 20150901,0
213 | 20150917,0
214 | 20151021,0
215 | 20150916,0
216 | 20151006,2
217 | 20150808,1
218 | 20150809,1
219 | 20150912,1
220 | 20150722,0
221 | 20151015,0
222 | 20150801,1
223 | 20150821,0
224 | 20150728,0
225 | 20151001,2
226 | 20151022,0
227 | 20150905,2
228 | 20150915,0
229 | 20150830,1
230 | 20151002,2
231 | 20151011,1
232 | 20150816,1
233 | 20150807,0
234 | 20151016,0
235 | 20150903,2
236 | 20150914,0
237 | 20151014,0
238 | 20151004,2
239 | 20150829,1
240 | 20150720,0
241 | 20150930,0
242 | 20150806,0
243 | 20151024,1
244 | 20150902,0
245 | 20150817,0
246 | 20150730,0
247 | 20150921,0
248 | 20151009,0
249 | 20150810,0
250 | 20150920,1
251 | 20150724,0
252 | 20150805,0
253 | 20150824,0
254 | 20150822,1
255 | 20150922,0
256 | 20150909,0
257 | 20150731,0
258 | 20150812,0
259 | 20150727,0
260 | 20151012,0
261 | 20150831,0
262 | 20150928,0
263 | 20150729,0
264 | 20150906,0
265 | 20150919,1
266 | 20150814,0
267 | 20150726,1
268 | 20151005,2
269 | 20150827,0
270 | 20150823,1
271 | 20150923,0
272 | 20151017,1
273 | 20150904,2
274 | 20151027,0
275 | 20150927,2
276 | 20150820,0
277 | 20150804,0
278 | 20150918,0
279 | 20150826,0
280 | 20151018,1
281 | 20150819,0
282 | 20151026,0
283 | 20151023,0
284 | 20150802,1
285 | 20151020,0
286 | 20150723,0
287 | 20150818,0
288 | 20151013,0
289 | 20151010,0
290 | 20151007,2
291 | 20150811,0
292 | 20150925,0
293 | 20150926,1
294 | 20150803,0
295 | 20150908,0
296 | 20151008,0
297 | 20151019,0
298 | 20150725,1
299 | 20150813,0
300 | 20150828,0
301 | 20150910,0
302 | 20151103,0
303 | 20160104,0
304 | 20160103,2
305 | 20151106,0
306 | 20151127,0
307 | 20151219,1
308 | 20151128,1
309 | 20160121,0
310 | 20151110,0
311 | 20151109,0
312 | 20160124,1
313 | 20151107,1
314 | 20151031,1
315 | 20151210,0
316 | 20151115,1
317 | 20151123,0
318 | 20160204,0
319 | 20151126,0
320 | 20151229,0
321 | 20151118,0
322 | 20160128,0
323 | 20151101,1
324 | 20151028,0
325 | 20151224,0
326 | 20151207,0
327 | 20160118,0
328 | 20151102,0
329 | 20151112,0
330 | 20151216,0
331 | 20160108,0
332 | 20160127,0
333 | 20151120,0
334 | 20151121,1
335 | 20151203,0
336 | 20160201,0
337 | 20151108,1
338 | 20160126,0
339 | 20160110,1
340 | 20160114,0
341 | 20151105,0
342 | 20151030,0
343 | 20160102,2
344 | 20160105,0
345 | 20151220,1
346 | 20151221,0
347 | 20151029,0
348 | 20151201,0
349 | 20151125,0
350 | 20160202,0
351 | 20160117,1
352 | 20160109,1
353 | 20160119,0
354 | 20160129,0
355 | 20151122,1
356 | 20151213,1
357 | 20151215,0
358 | 20151223,0
359 | 20151209,0
360 | 20160111,0
361 | 20160113,0
362 | 20151119,0
363 | 20151202,0
364 | 20151111,0
365 | 20151225,0
366 | 20151231,0
367 | 20151117,0
368 | 20160107,0
369 | 20151204,0
370 | 20160115,0
371 | 20160131,1
372 | 20151116,0
373 | 20151129,1
374 | 20151218,0
375 | 20151208,0
376 | 20160122,0
377 | 20151217,0
378 | 20151230,0
379 | 20151114,1
380 | 20160203,0
381 | 20160123,1
382 | 20160112,0
383 | 20151222,0
384 | 20151104,0
385 | 20151124,0
386 | 20151226,1
387 | 20151130,0
388 | 20151206,1
389 | 20151205,1
390 | 20151228,0
391 | 20160130,1
392 | 20160106,0
393 | 20160116,1
394 | 20151214,0
395 | 20160101,2
396 | 20151211,0
397 | 20151212,1
398 | 20151113,0
399 | 20160125,0
400 | 20160120,0
401 | 20151227,1
402 | 20160312,1
403 | 20160505,0
404 | 20160307,0
405 | 20160324,0
406 | 20160221,1
407 | 20160226,0
408 | 20160318,0
409 | 20160214,0
410 | 20160328,0
411 | 20160212,2
412 | 20160308,0
413 | 20160325,0
414 | 20160403,1
415 | 20160418,0
416 | 20160215,0
417 | 20160216,0
418 | 20160321,0
419 | 20160428,0
420 | 20160501,2
421 | 20160314,0
422 | 20160310,0
423 | 20160417,1
424 | 20160306,1
425 | 20160229,0
426 | 20160401,0
427 | 20160503,0
428 | 20160316,0
429 | 20160419,0
430 | 20160420,0
431 | 20160504,0
432 | 20160207,2
433 | 20160410,1
434 | 20160217,0
435 | 20160416,1
436 | 20160219,0
437 | 20160225,0
438 | 20160326,1
439 | 20160424,1
440 | 20160510,0
441 | 20160423,1
442 | 20160512,0
443 | 20160224,0
444 | 20160313,1
445 | 20160228,1
446 | 20160220,1
447 | 20160413,0
448 | 20160422,0
449 | 20160427,0
450 | 20160320,1
451 | 20160425,0
452 | 20160331,0
453 | 20160317,0
454 | 20160208,2
455 | 20160412,0
456 | 20160421,0
457 | 20160415,0
458 | 20160409,1
459 | 20160305,1
460 | 20160302,0
461 | 20160309,0
462 | 20160211,2
463 | 20160205,0
464 | 20160227,1
465 | 20160206,0
466 | 20160329,0
467 | 20160507,1
468 | 20160323,0
469 | 20160223,0
470 | 20160311,0
471 | 20160405,0
472 | 20160430,1
473 | 20160218,0
474 | 20160327,1
475 | 20160315,0
476 | 20160209,2
477 | 20160330,0
478 | 20160304,0
479 | 20160411,0
480 | 20160210,2
481 | 20160513,0
482 | 20160514,1
483 | 20160408,0
484 | 20160319,1
485 | 20160429,0
486 | 20160502,2
487 | 20160303,0
488 | 20160414,0
489 | 20160407,0
490 | 20160426,0
491 | 20160404,2
492 | 20160222,0
493 | 20160508,1
494 | 20160506,0
495 | 20160509,0
496 | 20160301,0
497 | 20160406,0
498 | 20160511,0
499 | 20160402,1
500 | 20160322,0
501 | 20160213,2
502 | 20160701,0
503 | 20160810,0
504 | 20160804,0
505 | 20160623,0
506 | 20160606,0
507 | 20160525,0
508 | 20160705,0
509 | 20160718,0
510 | 20160727,0
511 | 20160814,1
512 | 20160723,1
513 | 20160526,0
514 | 20160519,0
515 | 20160708,0
516 | 20160521,1
517 | 20160607,0
518 | 20160816,0
519 | 20160712,0
520 | 20160610,2
521 | 20160617,0
522 | 20160703,1
523 | 20160717,1
524 | 20160603,0
525 | 20160820,1
526 | 20160624,0
527 | 20160621,0
528 | 20160520,0
529 | 20160729,0
530 | 20160627,0
531 | 20160614,0
532 | 20160806,1
533 | 20160801,0
534 | 20160714,0
535 | 20160528,1
536 | 20160805,0
537 | 20160702,1
538 | 20160626,1
539 | 20160517,0
540 | 20160615,0
541 | 20160523,0
542 | 20160710,1
543 | 20160731,1
544 | 20160721,0
545 | 20160609,2
546 | 20160619,1
547 | 20160724,1
548 | 20160628,0
549 | 20160605,1
550 | 20160630,0
551 | 20160819,0
552 | 20160715,0
553 | 20160713,0
554 | 20160527,0
555 | 20160813,1
556 | 20160807,1
557 | 20160704,0
558 | 20160821,1
559 | 20160613,0
560 | 20160725,0
561 | 20160604,1
562 | 20160802,0
563 | 20160611,2
564 | 20160612,0
565 | 20160531,0
566 | 20160515,1
567 | 20160719,0
568 | 20160815,0
569 | 20160728,0
570 | 20160524,0
571 | 20160811,0
572 | 20160808,0
573 | 20160709,1
574 | 20160822,0
575 | 20160716,1
576 | 20160726,0
577 | 20160620,0
578 | 20160818,0
579 | 20160608,0
580 | 20160616,0
581 | 20160711,0
582 | 20160706,0
583 | 20160622,0
584 | 20160720,0
585 | 20160707,0
586 | 20160809,0
587 | 20160812,0
588 | 20160601,0
589 | 20160629,0
590 | 20160516,0
591 | 20160522,1
592 | 20160803,0
593 | 20160618,1
594 | 20160530,0
595 | 20160518,0
596 | 20160625,1
597 | 20160730,1
598 | 20160817,0
599 | 20160602,0
600 | 20160529,1
601 | 20160722,0
602 | 20160906,0
603 | 20160923,0
604 | 20161001,2
605 | 20161117,0
606 | 20161105,1
607 | 20161023,1
608 | 20161002,2
609 | 20161015,1
610 | 20160904,1
611 | 20160916,2
612 | 20161125,0
613 | 20160901,0
614 | 20161005,2
615 | 20161017,0
616 | 20161113,1
617 | 20160911,1
618 | 20161107,0
619 | 20161123,0
620 | 20161121,0
621 | 20161114,0
622 | 20161118,0
623 | 20161026,0
624 | 20161108,0
625 | 20161115,0
626 | 20160924,1
627 | 20161012,0
628 | 20160918,0
629 | 20160907,0
630 | 20161019,0
631 | 20160910,1
632 | 20161103,0
633 | 20161122,0
634 | 20160905,0
635 | 20161009,0
636 | 20160915,2
637 | 20161112,1
638 | 20161031,0
639 | 20160902,0
640 | 20161116,0
641 | 20160903,1
642 | 20161106,1
643 | 20160930,0
644 | 20161109,0
645 | 20160917,2
646 | 20161127,1
647 | 20161011,0
648 | 20160828,1
649 | 20160922,0
650 | 20161129,0
651 | 20161124,0
652 | 20160929,0
653 | 20161021,0
654 | 20161025,0
655 | 20161120,1
656 | 20161102,0
657 | 20161022,1
658 | 20161003,2
659 | 20161030,1
660 | 20161024,0
661 | 20161007,2
662 | 20160831,0
663 | 20160914,0
664 | 20160919,0
665 | 20160824,0
666 | 20160909,0
667 | 20160913,0
668 | 20160925,1
669 | 20160921,0
670 | 20161027,0
671 | 20160926,0
672 | 20161111,0
673 | 20161013,0
674 | 20160827,1
675 | 20160920,0
676 | 20160912,0
677 | 20161028,0
678 | 20161104,0
679 | 20160830,0
680 | 20161110,0
681 | 20161128,0
682 | 20160826,0
683 | 20161018,0
684 | 20161010,0
685 | 20161130,0
686 | 20161119,1
687 | 20161008,0
688 | 20161006,2
689 | 20161029,1
690 | 20160829,0
691 | 20161016,1
692 | 20160928,0
693 | 20160823,0
694 | 20161126,1
695 | 20161004,2
696 | 20160927,0
697 | 20161014,0
698 | 20161101,0
699 | 20161020,0
700 | 20160908,0
701 | 20160825,0
702 | 20161213,0
703 | 20161223,0
704 | 20161208,0
705 | 20161210,1
706 | 20161203,1
707 | 20161226,0
708 | 20161218,1
709 | 20161207,0
710 | 20161201,0
711 | 20161222,0
712 | 20161231,1
713 | 20161220,0
714 | 20161214,0
715 | 20161227,0
716 | 20161229,0
717 | 20161230,0
718 | 20161204,1
719 | 20161202,0
720 | 20161209,0
721 | 20161228,0
722 | 20161217,1
723 | 20161221,0
724 | 20161211,1
725 | 20161215,0
726 | 20161219,0
727 | 20161205,0
728 | 20161212,0
729 | 20161216,0
730 | 20161206,0
731 | 20161224,1
732 | 20161225,1
733 | 


--------------------------------------------------------------------------------
/dataset/yangzhong.csv:
--------------------------------------------------------------------------------
  1 | 2015-01-01,8,-4,晴,西南风,小于3级
  2 | 2015-01-02,13,0,多云转晴,东南风,小于3级
  3 | 2015-01-03,16,3,多云,南风,小于3级
  4 | 2015-01-04,15,4,晴转阴,东风,3-4级
  5 | 2015-01-05,10,2,小雨转雨夹雪,西北风,4-5级转3-4级
  6 | 2015-01-06,7,-1,阴转多云,北风,小于3级
  7 | 2015-01-07,7,-2,晴,西北风,小于3级
  8 | 2015-01-08,12,-1,晴,北风,小于3级
  9 | 2015-01-09,14,1,晴转多云,西南风,小于3级
 10 | 2015-01-10,14,1,晴,西南风,小于3级
 11 | 2015-01-11,10,1,多云,东北风,小于3级
 12 | 2015-01-12,8,3,阴,东风,小于3级
 13 | 2015-01-13,8,4,小到中雨转多云,东北风,小于3级
 14 | 2015-01-14,10,4,阴转多云,西北风,小于3级
 15 | 2015-01-15,11,3,多云,西南风,小于3级
 16 | 2015-01-16,7,-1,多云转晴,西北风,3-4级
 17 | 2015-01-17,9,-1,多云,东风,小于3级
 18 | 2015-01-18,11,0,多云,西风,小于3级
 19 | 2015-01-19,10,2,多云转阴,东风,小于3级
 20 | 2015-01-20,11,4,多云,东南风,小于3级转3-4级
 21 | 2015-01-21,7,0,多云,西北风,小于3级
 22 | 2015-01-22,11,-2,晴,东北风,小于3级
 23 | 2015-01-23,15,2,多云转阴,东南风,小于3级
 24 | 2015-01-24,11,7,阴转小雨,东风,3-4级
 25 | 2015-01-25,10,6,阴,西北风,小于3级转3-4级
 26 | 2015-01-26,6,2,阴,东北风,4-5级
 27 | 2015-01-27,3,0,小雪,东北风,4-5级转小于3级
 28 | 2015-01-28,2,-1,大雪,东北风,3-4级
 29 | 2015-01-29,3,-1,小雪转阴,西北风,小于3级
 30 | 2015-01-30,4,-2,多云,北风,小于3级
 31 | 2015-01-31,5,-1,阴转雨夹雪,东北风,小于3级
 32 | 2015-02-01,7,1,阴,东风,小于3级
 33 | 2015-02-02,7,2,阴转多云,东风,小于3级
 34 | 2015-02-03,7,2,阴转多云,西北风,3-4级
 35 | 2015-02-04,7,-2,多云转晴,西北风,3-4级
 36 | 2015-02-05,9,-2,晴转多云,西南风,小于3级
 37 | 2015-02-06,12,-1,多云,南风,小于3级转3-4级
 38 | 2015-02-07,6,0,多云转晴,西北风,4-5级
 39 | 2015-02-08,6,-4,晴,东北风,小于3级
 40 | 2015-02-09,15,0,晴,西南风,3-4级
 41 | 2015-02-10,15,0,晴,西南风,3-4级
 42 | 2015-02-11,15,0,晴,西南风,3-4级
 43 | 2015-02-12,15,0,晴,西南风,3-4级
 44 | 2015-02-13,19,3,晴转多云,东南风,小于3级
 45 | 2015-02-14,15,10,阵雨,东南风,小于3级
 46 | 2015-02-15,15,9,阴转多云,西北风,3-4级
 47 | 2015-02-16,14,3,多云转晴,西南风,小于3级
 48 | 2015-02-17,12,3,晴,东风,4-5级
 49 | 2015-02-18,11,1,多云,东北风,小于3级
 50 | 2015-02-19,10,4,阴转小到中雨,东风,3-4级转小于3级
 51 | 2015-02-20,11,8,小雨转阴,西风,3-4级
 52 | 2015-02-21,11,3,阴转多云,西北风,3-4级
 53 | 2015-02-22,10,4,小雨转阴,东风,小于3级
 54 | 2015-02-23,12,4,阴,东风,小于3级
 55 | 2015-02-24,10,6,小到中雨转小雨,东南风,3-4级
 56 | 2015-02-25,10,6,阴,暂无实况,小于3级转3-4级
 57 | 2015-02-26,6,3,阴转小到中雨,东风,小于3级
 58 | 2015-02-27,4,2,中雨,东北风,4-5级
 59 | 2015-02-28,9,1,阴转多云,西北风,3-4级
 60 | 2015-03-01,13,2,晴转多云,南风,3-4级
 61 | 2015-03-02,12,6,小雨转多云,东南风,3-4级转4-5级
 62 | 2015-03-03,6,1,多云,东北风,4-5级转3-4级
 63 | 2015-03-04,5,1,阴转雨夹雪,东风,小于3级
 64 | 2015-03-05,8,2,雨夹雪转阴,东北风,小于3级
 65 | 2015-03-06,12,3,多云,东南风,小于3级
 66 | 2015-03-07,10,5,阴转小雨,东风,3-4级
 67 | 2015-03-08,9,4,阴转多云,北风,小于3级转4-5级
 68 | 2015-03-09,7,0,多云,东北风,4-5级转3-4级
 69 | 2015-03-10,11,-1,多云,东南风,小于3级
 70 | 2015-03-11,16,2,晴,南风,小于3级
 71 | 2015-03-12,19,6,晴转多云,南风,3-4级转小于3级
 72 | 2015-03-13,15,7,多云转小雨,东北风,3-4级
 73 | 2015-03-14,16,8,阴,东风,3-4级
 74 | 2015-03-15,17,10,阴转多云,东风,小于3级
 75 | 2015-03-16,16,11,阴转中雨,东南风,小于3级
 76 | 2015-03-17,14,12,大雨转小到中雨,东风,3-4级
 77 | 2015-03-18,15,9,小雨转阴,北风,3-4级
 78 | 2015-03-19,14,9,小雨,暂无实况,3-4级
 79 | 2015-03-20,20,9,阴转多云,西风,小于3级
 80 | 2015-03-21,19,8,多云,西南风,小于3级
 81 | 2015-03-22,16,7,多云转阴,东风,4-5级转3-4级
 82 | 2015-03-23,15,6,多云,东风,小于3级
 83 | 2015-03-24,17,7,多云,东风,3-4级
 84 | 2015-03-25,15,6,多云转阴,东风,3-4级
 85 | 2015-03-26,16,9,小雨转多云,东风,3-4级
 86 | 2015-03-27,20,7,多云转晴,东南风,3-4级转小于3级
 87 | 2015-03-28,23,10,多云,南风,小于3级
 88 | 2015-03-29,20,14,阴转阵雨,东风,3-4级
 89 | 2015-03-30,26,16,多云,东南风,3-4级
 90 | 2015-03-31,23,17,多云转阴,东风,3-4级
 91 | 2015-04-01,25,14,阴转中雨,东风,5-6级转4-5级
 92 | 2015-04-02,19,11,阴,西北风,3-4级转4-5级
 93 | 2015-04-03,14,10,雷阵雨,东风,3-4级转小于3级
 94 | 2015-04-04,13,10,小雨转小到中雨,西北风,小于3级
 95 | 2015-04-05,11,7,中雨,北风,4-5级转3-4级
 96 | 2015-04-06,7,4,小雨,东北风,4-5级转3-4级
 97 | 2015-04-07,12,2,阴转多云,西风,小于3级
 98 | 2015-04-08,15,3,多云,东风,3-4级
 99 | 2015-04-09,17,6,晴转多云,东风,小于3级
100 | 2015-04-10,19,7,多云,北风,3-4级
101 | 2015-04-11,21,9,多云转雷阵雨,东风,3-4级
102 | 2015-04-12,13,7,多云转雷阵雨,西北风,5-6级
103 | 2015-04-13,19,6,阴转晴,西风,5-6级转3-4级
104 | 2015-04-14,23,10,晴,西南风,4-5级
105 | 2015-04-15,21,13,阴转多云,西南风,3-4级
106 | 2015-04-16,22,10,晴转多云,东南风,4-5级
107 | 2015-04-17,26,15,阴,东南风,4-5级转3-4级
108 | 2015-04-18,24,16,阴转雷阵雨,北风,3-4级
109 | 2015-04-19,17,10,中雨转阵雨,北风,4-5级转3-4级
110 | 2015-04-20,20,8,多云,暂无实况,小于3级
111 | 2015-04-21,24,8,多云转晴,南风,小于3级
112 | 2015-04-22,26,12,多云,东南风,3-4级
113 | 2015-04-23,27,13,多云,东南风,小于3级
114 | 2015-04-24,27,13,多云转晴,西南风,小于3级
115 | 2015-04-25,28,14,晴,暂无实况,小于3级
116 | 2015-04-26,28,15,晴转多云,东南风,4-5级
117 | 2015-04-27,29,18,阴转雷阵雨,南风,小于3级
118 | 2015-04-28,22,17,雷阵雨转多云,暂无实况,3-4级
119 | 2015-04-29,25,14,多云,西风,小于3级
120 | 2015-04-30,27,14,晴转多云,东风,小于3级
121 | 2015-05-01,20,16,小到中雨转中到大雨,东风,小于3级
122 | 2015-05-02,25,15,阴转多云,西北风,3-4级
123 | 2015-05-03,24,15,多云,暂无实况,小于3级
124 | 2015-05-04,24,12,阴转多云,东南风,3-4级转小于3级
125 | 2015-05-05,23,15,阴,东南风,3-4级
126 | 2015-05-06,27,16,多云,东南风,小于3级
127 | 2015-05-07,27,17,多云,北风,3-4级
128 | 2015-05-08,26,15,多云,西南风,3-4级
129 | 2015-05-09,26,15,多云,西南风,3-4级
130 | 2015-05-10,26,15,多云,西南风,3-4级
131 | 2015-05-11,26,15,多云,西南风,3-4级
132 | 2015-05-12,26,15,多云,西南风,3-4级
133 | 2015-05-13,29,20,晴~多云,东南风,3-4级
134 | 2015-05-14,31,22,多云~中雨,西南风,3-4级
135 | 2015-05-15,25,16,小到中雨~阴,东北风,3-4级
136 | 2015-05-16,24,15,多云,东南风,3-4级
137 | 2015-05-17,26,18,多云~雷雨,东南风,3-4级
138 | 2015-05-18,25,17,小雨~多云,东南风,3-4级
139 | 2015-05-19,29,19,多云,东南风,3-4级
140 | 2015-05-20,27,15,多云,东南风,3-4级
141 | 2015-05-21,25,15,晴~多云,东南风,3-4级
142 | 2015-05-22,28,18,晴,东北风,3-4级
143 | 2015-05-23,26,18,阴~,东北风,3-4级
144 | 2015-05-24,27,19,多云,东北风,3-4级
145 | 2015-05-25,29,19,晴~,东北风,3-4级
146 | 2015-05-26,27,20,多云~,东南风,3-4级
147 | 2015-05-27,22,17,阵雨,东北风,3-4级
148 | 2015-05-28,27,20,阴,东南风,3-4级
149 | 2015-05-29,23,20,中到大雨~中雨,东北风,3-4级
150 | 2015-05-30,26,19,多云~晴,东北风,3-4级
151 | 2015-05-31,27,19,多云,东南风,3-4级
152 | 2015-06-01,31,23,多云~雷雨,东南风,3-4级
153 | 2015-06-02,25,19,暴雨~小到中雨,东北风,3-4级
154 | 2015-06-03,26,16,阴~晴,东南风,3-4级
155 | 2015-06-04,26,19,多云,西北风,3-4级
156 | 2015-06-05,28,19,晴,西南风,3-4级
157 | 2015-06-06,32,21,多云,东南风,3-4级
158 | 2015-06-07,27,21,雷雨~大雨,东北风,3-4级
159 | 2015-06-08,27,20,阴,西南风,3-4级
160 | 2015-06-09,29,22,阴,东南风,3-4级
161 | 2015-06-10,28,21,阴~多云,东南风,3-4级
162 | 2015-06-11,33,22,多云,西南风,3-4级
163 | 2015-06-12,34,23,晴~多云,西南风,3-4级
164 | 2015-06-13,33,25,多云,西南风,3-4级
165 | 2015-06-14,32,23,多云~阴,东北风,3-4级
166 | 2015-06-15,28,21,雷雨,东南风,3-4级
167 | 2015-06-16,21,18,大雨~暴雨,东北风,3-4级
168 | 2015-06-17,22,18,小雨~阴,东北风,3-4级
169 | 2015-06-18,28,19,多云,东南风,3-4级
170 | 2015-06-19,29,21,多云,东南风,3-4级
171 | 2015-06-20,30,21,多云~阴,东北风,3-4级
172 | 2015-06-21,26,21,阵雨,东南风,3-4级
173 | 2015-06-22,26,21,阴,东南风,3-4级
174 | 2015-06-23,25,23,阵雨~小雨,东南风,3-4级
175 | 2015-06-24,26,22,小雨~小到中雨,东北风,3-4级
176 | 2015-06-25,30,23,阴~中雨,西北风,3-4级
177 | 2015-06-26,26,21,雷雨~大雨,东南风,3-4级
178 | 2015-06-27,23,19,暴雨,东南风,3-4级
179 | 2015-06-28,24,21,中到大雨~小雨,东南风,3-4级
180 | 2015-06-29,24,21,大雨~雷雨,东南风,3-4级
181 | 2015-06-30,27,21,中到大雨~小雨,东南风,3-4级
182 | 2015-07-01,29,21,多云~晴,西南风,3-4级
183 | 2015-07-02,32,23,多云,西南风,3-4级
184 | 2015-07-03,30,21,多云,东南风,3-4级
185 | 2015-07-04,28,19,阴,东南风,3-4级
186 | 2015-07-05,28,19,阴,东南风,3-4级
187 | 2015-07-06,22,18,小雨~中到大雨,东北风,3-4级
188 | 2015-07-07,24,19,雷雨,东北风,3-4级
189 | 2015-07-08,24,20,小到中雨~小雨,东北风,3-4级
190 | 2015-07-09,27,22,阴,东北风,3-4级
191 | 2015-07-10,31,24,阴~阵雨,东北风,3-4级
192 | 2015-07-11,27,23,大雨~大到暴雨,东北风,3-4级
193 | 2015-07-12,29,23,阴~多云,东南风,3-4级
194 | 2015-07-13,35,25,多云,东南风,3-4级
195 | 2015-07-14,35,25,晴~多云,东南风,3-4级
196 | 2015-07-15,32,24,多云,东北风,3-4级
197 | 2015-07-16,29,23,雷阵雨~小到中雨,东南风,3-4级
198 | 2015-07-17,25,22,中雨,东南风,3-4级
199 | 2015-07-18,27,23,雷阵雨,东南风,3-4级
200 | 2015-07-19,26,24,大雨~小雨,东南风,3-4级
201 | 2015-07-20,28,24,阴,东南风,3-4级
202 | 2015-07-21,30,24,多云,东南风,3-4级
203 | 2015-07-22,31,25,多云~阴,东南风,3-4级
204 | 2015-07-23,31,25,雷阵雨,东南风,3-4级
205 | 2015-07-24,28,24,中雨,西南风,3-4级
206 | 2015-07-25,29,25,中雨~小雨,东南风,3-4级
207 | 2015-07-26,32,26,雷阵雨~多云,西南风,3-4级
208 | 2015-07-27,35,27,多云,西南风,3-4级
209 | 2015-07-28,36,27,多云~晴,西南风,3-4级
210 | 2015-07-29,37,28,多云,西南风,3-4级
211 | 2015-07-30,35,27,晴~多云,西南风,3-4级
212 | 2015-07-31,35,27,晴,西北风,3-4级
213 | 2015-08-01,35,27,阴~多云,东南风,3-4级
214 | 2015-08-02,37,28,晴,西南风,3-4级
215 | 2015-08-03,37,28,晴,西南风,3-4级
216 | 2015-08-04,37,28,晴,西南风,3-4级
217 | 2015-08-05,37,28,多云,东南风,3-4级
218 | 2015-08-06,35,26,多云,东南风,3-4级
219 | 2015-08-07,34,26,多云,东南风,3-4级
220 | 2015-08-08,33,26,多云~雷阵雨,东北风,3-4级
221 | 2015-08-09,30,25,雷阵雨~小到中雨,东北风,3-4级
222 | 2015-08-10,28,24,暴雨~小到中雨,东南风,3-4级
223 | 2015-08-11,28,23,阴~多云,西北风,3-4级
224 | 2015-08-12,31,23,多云,东南风,3-4级
225 | 2015-08-13,30,23,多云,东南风,3-4级
226 | 2015-08-14,32,23,多云,东南风,3-4级
227 | 2015-08-15,30,24,阴,东南风,3-4级
228 | 2015-08-16,30,23,多云~晴,东南风,3-4级
229 | 2015-08-17,31,24,多云,东南风,3-4级
230 | 2015-08-18,30,23,多云,东北风,3-4级
231 | 2015-08-19,30,24,阴~大雨,东南风,3-4级
232 | 2015-08-20,26,23,阵雨~阴,无持续风向,3-4级
233 | 2015-08-21,29,23,阴~多云,东北风,3-4级
234 | 2015-08-22,28,21,阴,东南风,3-4级
235 | 2015-08-23,28,21,阴,东北风,3-4级
236 | 2015-08-24,24,20,小到中雨~多云,西北风,3-4级
237 | 2015-08-25,29,21,多云,东北风,3-4级
238 | 2015-08-26,30,22,多云,西南风,3-4级
239 | 2015-08-27,30,22,阵雨~多云,东南风,3-4级
240 | 2015-08-28,30,22,多云,东南风,3-4级
241 | 2015-08-29,31,22,多云,东南风,3-4级
242 | 2015-08-30,31,22,多云,东南风,3-4级
243 | 2015-08-31,30,22,多云,东北风,3-4级
244 | 2015-09-01,29,22,雷阵雨~多云,东北风,3-4级
245 | 2015-09-02,30,22,多云,西北风,3-4级
246 | 2015-09-03,29,22,多云,东南风,3-4级
247 | 2015-09-04,30,23,阴~中雨,东北风,3-4级
248 | 2015-09-05,28,22,阴~多云,西北风,3-4级
249 | 2015-09-06,28,21,多云,东北风,3-4级
250 | 2015-09-07,28,21,多云,东北风,3-4级
251 | 2015-09-08,28,20,多云,东北风,3-4级
252 | 2015-09-09,28,19,晴~多云,东北风,3-4级
253 | 2015-09-10,28,20,晴~多云,东南风,3-4级
254 | 2015-09-11,27,20,多云~阵雨,西北风,3-4级
255 | 2015-09-12,24,16,阴~多云,西北风,3-4级
256 | 2015-09-13,25,17,多云,东北风,3-4级
257 | 2015-09-14,26,20,多云,东南风,3-4级
258 | 2015-09-15,26,18,阴~多云,东南风,3-4级
259 | 2015-09-16,27,18,多云~晴,东北风,3-4级
260 | 2015-09-17,26,18,多云,西北风,3-4级
261 | 2015-09-18,27,19,多云,东北风,3-4级
262 | 2015-09-19,28,20,多云,东南风,3-4级
263 | 2015-09-20,26,19,多云~晴,东南风,3-4级
264 | 2015-09-21,27,20,多云,东南风,3-4级
265 | 2015-09-22,26,21,阵雨~小雨,西南风,3-4级
266 | 2015-09-23,26,19,多云,西北风,3-4级
267 | 2015-09-24,26,20,多云~小雨,东北风,3-4级
268 | 2015-09-25,27,19,阴~多云,东北风,3-4级
269 | 2015-09-26,28,18,多云~晴,东北风,3-4级
270 | 2015-09-27,27,18,晴,东北风,3-4级
271 | 2015-09-28,27,20,多云~阵雨,东北风,3-4级
272 | 2015-09-29,25,21,阵雨,东北风,3-4级
273 | 2015-09-30,26,19,阵雨~中到大雨,东南风,3-4级
274 | 2015-10-01,23,14,多云~晴,西北风,3-4级
275 | 2015-10-02,24,15,晴,东南风,3-4级
276 | 2015-10-03,25,17,多云,东南风,3-4级
277 | 2015-10-04,25,19,多云~小雨,西南风,3-4级
278 | 2015-10-05,23,18,阴~多云,东南风,3-4级
279 | 2015-10-06,24,19,阴~阵雨,东南风,3-4级
280 | 2015-10-07,22,18,阵雨,东北风,3-4级
281 | 2015-10-08,19,15,阴~多云,东北风,3-4级
282 | 2015-10-09,22,12,多云~晴,西南风,3-4级
283 | 2015-10-10,23,10,晴~多云,西北风,3-4级
284 | 2015-10-11,20,10,晴,东南风,3-4级
285 | 2015-10-12,22,11,晴,东南风,3-4级
286 | 2015-10-13,24,13,晴~多云,东南风,3-4级
287 | 2015-10-14,24,13,多云,东南风,3-4级
288 | 2015-10-15,25,14,晴,西南风,3-4级
289 | 2015-10-16,26,15,晴,东南风,3-4级
290 | 2015-10-17,26,14,晴,东南风,3-4级
291 | 2015-10-18,25,14,晴~多云,东北风,3-4级
292 | 2015-10-19,24,15,多云,东南风,3-4级
293 | 2015-10-20,25,16,多云,东南风,3-4级
294 | 2015-10-21,25,16,多云,东北风,3-4级
295 | 2015-10-22,25,16,多云,东北风,3-4级
296 | 2015-10-23,25,17,多云,东北风,3-4级
297 | 2015-10-24,25,17,多云,东北风,3-4级
298 | 2015-10-25,23,16,多云~阴,东南风,3-4级
299 | 2015-10-26,24,14,小雨,西北风,3-4级
300 | 2015-10-27,19,11,多云,西南风,3-4级
301 | 2015-10-28,20,12,阴,东南风,3-4级
302 | 2015-10-29,18,12,阴~多云,西北风,3-4级
303 | 2015-10-30,18,10,多云,东北风,3-4级
304 | 2015-10-31,17,11,多云~阵雨,西北风,3-4级
305 | 2015-11-01,17,8,多云~晴,西南风,3-4级
306 | 2015-11-02,18,8,晴,东南风,3-4级
307 | 2015-11-03,20,11,多云,东南风,3-4级
308 | 2015-11-04,21,14,阴,东南风,3-4级
309 | 2015-11-05,23,16,阴~小雨,东南风,3-4级
310 | 2015-11-06,26,19,阵雨,东南风,3-4级
311 | 2015-11-07,17,12,阵雨~小雨,西北风,3-4级
312 | 2015-11-08,13,9,阵雨~多云,西北风,3-4级
313 | 2015-11-09,13,8,多云,东北风,3-4级
314 | 2015-11-10,15,9,多云,东南风,3-4级
315 | 2015-11-11,16,10,阴,东北风,3-4级
316 | 2015-11-12,16,13,小雨,西南风,3-4级
317 | 2015-11-13,16,12,阴,西北风,3-4级
318 | 2015-11-14,17,10,多云,东南风,3-4级
319 | 2015-11-15,18,11,阴~中雨,东风,3-4级
320 | 2015-11-16,17,13,阴~小雨,东北风,3-4级
321 | 2015-11-17,15,12,小雨,东北风,3-4级
322 | 2015-11-18,15,11,小雨~小到中雨,东北风,3-4级
323 | 2015-11-19,14,11,小雨~阴,东北风,3-4级
324 | 2015-11-20,16,11,阴,东南风,3-4级
325 | 2015-11-21,16,13,阴~阵雨,东北风,3-4级
326 | 2015-11-22,17,14,小雨,东北风,3-4级
327 | 2015-11-23,16,10,小雨,东北风,3-4级
328 | 2015-11-24,11,3,小雨~小到中雨,西北风,3-4级
329 | 2015-11-25,5,0,多云~晴,西北风,3-4级
330 | 2015-11-26,3,-4,晴,西北风,3-4级
331 | 2015-11-27,5,-2,晴~多云,东南风,3-4级
332 | 2015-11-28,12,3,多云~阴,东北风,3-4级
333 | 2015-11-29,16,6,多云~晴,东南风,3-4级
334 | 2015-11-30,15,5,多云,东南风,3-4级
335 | 2015-12-01,17,10,多云~阴,西北风,3-4级
336 | 2015-12-02,12,3,多云~晴,西北风,3-4级
337 | 2015-12-03,10,0,晴,西北风,3-4级
338 | 2015-12-04,9,2,晴~多云,西南风,3-4级
339 | 2015-12-05,8,4,阵雨,西南风,3-4级
340 | 2015-12-06,10,2,多云,东南风,3-4级
341 | 2015-12-07,12,3,晴,东南风,3-4级
342 | 2015-12-08,13,6,多云,东南风,3-4级
343 | 2015-12-09,12,8,小雨,东北风,3-4级
344 | 2015-12-10,12,7,阴~多云,西北风,3-4级
345 | 2015-12-11,10,1,多云,东北风,3-4级
346 | 2015-12-12,10,2,晴~多云,东北风,3-4级
347 | 2015-12-13,9,3,多云~阴,西北风,3-4级
348 | 2015-12-14,10,4,阴,西北风,3-4级
349 | 2015-12-15,10,0,多云~晴,西北风,3-4级
350 | 2015-12-16,6,-3,晴,西北风,3-4级
351 | 2015-12-17,5,-3,晴,西南风,3-4级
352 | 2015-12-18,6,-2,多云,东南风,3-4级
353 | 2015-12-19,9,2,阴~小雨,东南风,3-4级
354 | 2015-12-20,8,2,小雨~阴,西北风,3-4级
355 | 2015-12-21,8,0,多云~阴,东南风,3-4级
356 | 2015-12-22,11,6,阴,西北风,3-4级
357 | 2015-12-23,10,7,阴~小雨,东北风,3-4级
358 | 2015-12-24,10,4,小雨~阴,西北风,3-4级
359 | 2015-12-25,10,2,多云~晴,西南风,3-4级
360 | 2015-12-26,14,4,晴~多云,西南风,3-4级
361 | 2015-12-27,9,1,阴~多云,东北风,3-4级
362 | 2015-12-28,6,-1,多云,东北风,3-4级
363 | 2015-12-29,9,0,多云,东南风,3-4级
364 | 2015-12-30,11,0,晴,西北风,3-4级
365 | 2015-12-31,9,-1,晴,东北风,3-4级
366 | 2016-01-01,11,2,多云,东南风,3-4级
367 | 2016-01-02,15,7,多云~阴,东南风,3-4级
368 | 2016-01-03,16,7,多云~阴,东南风,3-4级
369 | 2016-01-04,13,7,小雨~小到中雨,东北风,3-4级
370 | 2016-01-05,10,5,小雨~阴,东北风,3-4级
371 | 2016-01-06,8,3,阴~多云,西北风,3-4级
372 | 2016-01-07,6,-1,阴~多云,西北风,3-4级
373 | 2016-01-08,7,-1,晴~多云,东北风,3-4级
374 | 2016-01-09,9,-1,多云~阴,东南风,3-4级
375 | 2016-01-10,10,4,小雨,东北风,3-4级
376 | 2016-01-11,8,1,雨夹雪,东北风,3-4级
377 | 2016-01-12,3,-3,小雪~阴,西北风,3-4级
378 | 2016-01-13,4,-2,多云,西南风,3-4级
379 | 2016-01-14,7,-2,多云,西南风,3-4级
380 | 2016-01-15,8,-2,晴,东南风,3-4级
381 | 2016-01-16,10,1,多云~小雨,东北风,3-4级
382 | 2016-01-17,8,0,小雨~多云,东北风,3-4级
383 | 2016-01-18,3,-4,晴,西北风,3-4级
384 | 2016-01-19,3,-4,多云,东北风,3-4级
385 | 2016-01-20,3,-2,阴~中到大雪,东北风,3-4级
386 | 2016-01-21,2,-2,雨夹雪~阴,东北风,3-4级
387 | 2016-01-22,3,-4,小雪,东北风,3-4级
388 | 2016-01-23,-3,-10,阴~多云,西北风,3-4级
389 | 2016-01-24,-6,-10,晴,西北风,3-4级
390 | 2016-01-25,2,-8,晴,西南风,3-4级
391 | 2016-01-26,6,-4,多云~阴,西南风,3-4级
392 | 2016-01-27,8,-1,多云~小雨,东南风,3-4级
393 | 2016-01-28,6,3,小雨,东北风,3-4级
394 | 2016-01-29,5,1,小雨~阴,东北风,3-4级
395 | 2016-01-30,5,-1,阴~多云,东北风,3-4级
396 | 2016-01-31,4,-2,雨夹雪~阴,东北风,3-4级
397 | 2016-02-01,2,-5,多云,西北风,3-4级
398 | 2016-02-02,5,-4,多云,东北风,3-4级
399 | 2016-02-03,7,-3,晴~多云,东北风,3-4级
400 | 2016-02-04,7,-2,多云,东南风,3-4级
401 | 2016-02-05,9,-2,多云~阴,西北风,3-4级
402 | 2016-02-06,7,-3,晴,西北风,3-4级
403 | 2016-02-07,10,-2,多云~晴,西北风,3-4级
404 | 2016-02-08,15,2,晴,西南风,3-4级
405 | 2016-02-09,17,2,晴~多云,东南风,3-4级
406 | 2016-02-10,20,7,多云~阴,东南风,3-4级
407 | 2016-02-11,18,11,雷阵雨~阴,东南风,3-4级
408 | 2016-02-12,18,11,阵雨~中雨,东北风,3-4级
409 | 2016-02-13,14,1,小雨~阴,西北风,3-4级
410 | 2016-02-14,6,-3,阴~多云,西北风,3-4级
411 | 2016-02-15,7,-2,晴,西北风,3-4级
412 | 2016-02-16,10,0,晴,西北风,3-4级
413 | 2016-02-17,12,2,多云,东南风,3-4级
414 | 2016-02-18,16,3,晴~多云,东南风,3-4级
415 | 2016-02-19,13,3,阴,东南风,3-4级
416 | 2016-02-20,11,-1,多云~晴,东南风,3-4级
417 | 2016-02-21,10,2,阴~小到中雨,东南风,3-4级
418 | 2016-02-22,7,4,小雨~阴,东北风,3-4级
419 | 2016-02-23,11,1,阴~多云,东北风,3-4级
420 | 2016-02-24,9,0,晴,东南风,3-4级
421 | 2016-02-25,12,2,多云,东南风,3-4级
422 | 2016-02-26,17,4,多云,西南风,3-4级
423 | 2016-02-27,19,8,晴~多云,西南风,3-4级
424 | 2016-02-28,17,2,阴~多云,东北风,3-4级
425 | 2016-02-29,10,-1,晴,东北风,3-4级
426 | 2016-03-01,13,1,晴~多云,东南风,3-4级
427 | 2016-03-02,16,7,多云,东南风,3-4级
428 | 2016-03-03,16,9,阵雨~阴,东南风,3-4级
429 | 2016-03-04,19,12,阵雨~阴,东北风,3-4级
430 | 2016-03-05,21,11,阴~多云,东北风,3-4级
431 | 2016-03-06,19,10,多云,东南风,3-4级
432 | 2016-03-07,16,6,多云~小雨,东北风,3-4级
433 | 2016-03-08,9,1,中到大雨~小雨,东北风,3-4级
434 | 2016-03-09,7,1,阴,东北风,3-4级
435 | 2016-03-10,8,-1,多云~晴,东北风,3-4级
436 | 2016-03-11,11,1,多云~晴,西南风,3-4级
437 | 2016-03-12,13,4,小雨,东南风,3-4级
438 | 2016-03-13,11,4,阴~多云,东北风,3-4级
439 | 2016-03-14,13,3,多云,东南风,3-4级
440 | 2016-03-15,14,4,多云,东南风,3-4级
441 | 2016-03-16,18,6,多云~阴,东南风,3-4级
442 | 2016-03-17,15,9,小雨,东北风,3-4级
443 | 2016-03-18,18,11,阴,西北风,3-4级
444 | 2016-03-19,19,7,多云,东南风,3-4级
445 | 2016-03-20,17,6,多云,东北风,3-4级
446 | 2016-03-21,17,6,多云,东北风,3-4级
447 | 2016-03-22,16,9,多云~阴,东南风,3-4级
448 | 2016-03-23,14,4,多云,东北风,3-4级
449 | 2016-03-24,12,3,多云,东北风,3-4级
450 | 2016-03-25,15,3,多云,东北风,3-4级
451 | 2016-03-26,15,5,阴~多云,西北风,3-4级
452 | 2016-03-27,18,7,晴,西风,3-4级
453 | 2016-03-28,20,9,多云,西南风,3-4级
454 | 2016-03-29,20,13,多云,西南风,3-4级
455 | 2016-03-30,21,12,阴~小雨,西南风,3-4级
456 | 2016-03-31,20,10,多云~晴,西南风,3-4级
457 | 2016-04-01,24,14,多云,东南风,3-4级
458 | 2016-04-02,19,13,雷阵雨~中雨,东北风,3-4级
459 | 2016-04-03,14,9,阵雨~阴,东北风,3-4级
460 | 2016-04-04,18,10,多云,东北风,3-4级
461 | 2016-04-05,16,11,阵雨~中雨,东南风,3-4级
462 | 2016-04-06,14,12,中到大雨~中雨,西北风,3-4级
463 | 2016-04-07,20,10,阴~多云,西南风,3-4级
464 | 2016-04-08,23,13,多云,东南风,3-4级
465 | 2016-04-09,25,13,多云,东南风,3-4级
466 | 2016-04-10,26,11,多云,东北风,3-4级
467 | 2016-04-11,21,12,多云~阴,东南风,3-4级
468 | 2016-04-12,23,14,多云,东南风,3-4级
469 | 2016-04-13,27,16,多云,东南风,3-4级
470 | 2016-04-14,27,15,多云,东南风,3-4级
471 | 2016-04-15,25,16,多云~小到中雨,东南风,3-4级
472 | 2016-04-16,20,12,中到大雨~阴,西北风,3-4级
473 | 2016-04-17,20,11,阴,东北风,3-4级
474 | 2016-04-18,22,12,多云,西南风,3-4级
475 | 2016-04-19,24,15,多云~雷阵雨,东南风,3-4级
476 | 2016-04-20,18,14,大雨~小雨,西北风,3-4级
477 | 2016-04-21,25,15,多云,西南风,3-4级
478 | 2016-04-22,27,16,多云~雷阵雨,东南风,3-4级
479 | 2016-04-23,18,13,小雨~阴,东南风,3-4级
480 | 2016-04-24,21,13,阴~多云,东北风,3-4级
481 | 2016-04-25,20,13,阴,东南风,3-4级
482 | 2016-04-26,17,12,小雨~小到中雨,东北风,3-4级
483 | 2016-04-27,17,11,阵雨~多云,西北风,3-4级
484 | 2016-04-28,23,12,多云,西北风,3-4级
485 | 2016-04-29,27,15,晴,西南风,3-4级
486 | 2016-04-30,29,17,晴~多云,东南风,3-4级
487 | 2016-05-01,30,20,多云~雷阵雨,东南风,3-4级
488 | 2016-05-02,26,16,中到大雨~小雨,东南风,3-4级
489 | 2016-05-03,25,14,晴,西南风,3-4级
490 | 2016-05-04,30,20,多云~阴,东南风,3-4级
491 | 2016-05-05,25,20,中雨~阴,西南风,3-4级
492 | 2016-05-06,28,17,多云,东北风,3-4级
493 | 2016-05-07,19,15,阴~小雨,东北风,3-4级
494 | 2016-05-08,17,13,阵雨,东南风,3-4级
495 | 2016-05-09,20,15,阵雨,东北风,3-4级
496 | 2016-05-10,21,13,多云~晴,西北风,3-4级
497 | 2016-05-11,27,16,晴,东南风,3-4级
498 | 2016-05-12,30,16,多云,东南风,3-4级
499 | 2016-05-13,20,15,阴~多云,东北风,3-4级
500 | 2016-05-14,26,16,多云~阴,东南风,3-4级
501 | 2016-05-15,18,13,中雨~多云,西北风,3-4级
502 | 2016-05-16,25,13,晴,东南风,3-4级
503 | 2016-05-17,26,14,多云,东南风,3-4级
504 | 2016-05-18,25,14,多云,东北风,3-4级
505 | 2016-05-19,27,17,多云,东南风,3-4级
506 | 2016-05-20,24,17,阴~小到中雨,东北风,3-4级
507 | 2016-05-21,19,16,中雨~小雨,东北风,3-4级
508 | 2016-05-22,19,15,小雨~阴,西北风,3-4级
509 | 2016-05-23,23,17,阴~小雨,西南风,3-4级
510 | 2016-05-24,26,17,阴~多云,西南风,3-4级
511 | 2016-05-25,29,21,多云~雷阵雨,西南风,3-4级
512 | 2016-05-26,26,20,雷阵雨~小到中雨,西南风,3-4级
513 | 2016-05-27,23,19,中雨~中到大雨,东北风,3-4级
514 | 2016-05-28,21,19,阵雨,东南风,3-4级
515 | 2016-05-29,24,17,阴~多云,东南风,3-4级
516 | 2016-05-30,27,20,多云~小雨,东南风,3-4级
517 | 2016-05-31,29,21,雷阵雨~大到暴雨,东北风,3-4级
518 | 2016-06-01,21,16,小雨~阴,东南风,3-4级
519 | 2016-06-02,22,18,多云~阵雨,东北风,3-4级
520 | 2016-06-03,23,18,阵雨,东北风,3-4级
521 | 2016-06-04,22,16,阴~多云,西北风,3-4级
522 | 2016-06-05,26,18,多云,东南风,3-4级
523 | 2016-06-06,27,19,雷阵雨~中雨,东北风,3-4级
524 | 2016-06-07,26,21,雷阵雨,东北风,3-4级
525 | 2016-06-08,24,19,阵雨~多云,东北风,3-4级
526 | 2016-06-09,25,19,阴,东北风,3-4级
527 | 2016-06-10,27,22,多云,东北风,3-4级
528 | 2016-06-11,30,23,阴~雷阵雨,西北风,3-4级
529 | 2016-06-12,26,20,雷阵雨~多云,东南风,3-4级
530 | 2016-06-13,31,22,多云,东南风,3-4级
531 | 2016-06-14,33,25,多云~阴,西南风,3-4级
532 | 2016-06-15,30,22,雷阵雨~阴,西北风,3-4级
533 | 2016-06-16,30,20,多云~晴,西南风,3-4级
534 | 2016-06-17,32,23,晴~多云,东南风,3-4级
535 | 2016-06-18,33,25,多云~雷阵雨,西南风,3-4级
536 | 2016-06-19,32,25,雷阵雨~中雨,东南风,3-4级
537 | 2016-06-20,29,25,雷阵雨,西南风,3-4级
538 | 2016-06-21,29,24,小到中雨~小雨,东南风,3-4级
539 | 2016-06-22,31,25,大到暴雨~阴,西南风,3-4级
540 | 2016-06-23,33,27,雷阵雨~阴,西南风,3-4级
541 | 2016-06-24,27,21,中雨~小雨,西南风,3-4级
542 | 2016-06-25,27,20,阵雨~多云,西南风,3-4级
543 | 2016-06-26,29,21,多云~小雨,东南风,3-4级
544 | 2016-06-27,22,19,小到中雨~雷阵雨,东北风,3-4级
545 | 2016-06-28,25,21,小到中雨~大雨,西南风,3-4级
546 | 2016-06-29,28,22,阴~多云,东南风,3-4级
547 | 2016-06-30,33,25,多云~雷阵雨,东南风,3-4级
548 | 2016-07-01,26,23,大到暴雨,东南风,3-4级
549 | 2016-07-02,26,22,中到大雨~暴雨,东南风,3-4级
550 | 2016-07-03,25,22,暴雨~大雨,东南风,3-4级
551 | 2016-07-04,27,22,中雨~大雨,东南风,3-4级
552 | 2016-07-05,28,23,雷阵雨,3-4级
553 | 2016-07-06,28,25,中雨~小雨,3-4级
554 | 2016-07-07,30,24,雷阵雨~多云,3-4级
555 | 2016-07-08,32,24,阴~多云,3-4级
556 | 2016-07-09,31,24,多云~阴,3-4级
557 | 2016-07-10,28,24,雷阵雨~中雨,3-4级
558 | 2016-07-11,29,24,阵雨~阴,3-4级
559 | 2016-07-12,32,26,多云~阴,东北风,3-4级
560 | 2016-07-13,32,26,中雨~雷阵雨,西南风,3-4级
561 | 2016-07-14,29,26,中雨转雷阵雨,东风,3-4级
562 | 2016-07-15,33,22,中到大雨转小雨,西风,3-4级
563 | 2016-07-16,29,22,多云,东南风,3-4级
564 | 2016-07-17,32,24,多云转雷阵雨,东南风,3-4级
565 | 2016-07-18,32,26,阴,东风,3-4级
566 | 2016-07-19,32,26,雷阵雨,南风,3-4级
567 | 2016-07-20,36,28,晴,南风,3-4级
568 | 2016-07-21,37,29,多云,西风,3-4级
569 | 2016-07-22,37,29,晴,南风,3-4级
570 | 2016-07-23,38,30,晴,西南风,3-4级
571 | 2016-07-24,38,30,晴,西南风,3-4级
572 | 2016-07-25,38,29,晴,东南风,3-4级
573 | 2016-07-26,39,30,晴,东南风,3-4级
574 | 2016-07-27,38,30,多云,北风,3-4级
575 | 2016-07-28,38,29,晴,西南风,3-4级
576 | 2016-07-29,38,30,晴,东南风,3-4级
577 | 2016-07-30,37,28,晴,东风,3-4级
578 | 2016-07-31,36,27,多云,东南风,3-4级
579 | 2016-08-01,35,27,晴,东南风,2级
580 | 2016-08-02,35,27,雷阵雨,东南风,3级
581 | 2016-08-03,32,25,雷阵雨,西北风,1级
582 | 2016-08-04,33,26,雷阵雨,东南风,1级
583 | 2016-08-05,33,26,雷阵雨,东风,2级
584 | 2016-08-06,33,26,雷阵雨,东南风,2级
585 | 2016-08-07,34,26,多云,东风,2级
586 | 2016-08-08,35,26,雷阵雨,东风,3级
587 | 2016-08-09,34,27,多云,东风,3级
588 | 2016-08-10,33,27,阵雨,东风,3级
589 | 2016-08-11,35,27,多云,东风,2级
590 | 2016-08-12,35,27,多云,东风,3级
591 | 2016-08-13,36,28,多云,东风,3级
592 | 2016-08-14,35,27,晴,东风,2级
593 | 2016-08-15,35,27,晴,东风,3级
594 | 2016-08-16,34,26,多云,东风,2级
595 | 2016-08-17,34,26,多云,东北风,3级
596 | 2016-08-18,35,28,晴,东风,2级
597 | 2016-08-19,36,28,晴,东风,2级
598 | 2016-08-20,36,27,雷阵雨,东南风,微风
599 | 2016-08-21,31,26,雷阵雨,东南风,1级
600 | 2016-08-22,34,27,多云,东北风,2级
601 | 2016-08-23,34,24,多云,东北风,2级
602 | 2016-08-24,33,23,晴,东风,2级
603 | 2016-08-25,34,25,多云,北风,1级
604 | 2016-08-26,28,23,阵雨,北风,2级
605 | 2016-08-27,29,20,多云,北风,3级
606 | 2016-08-28,30,19,晴,北风,1级
607 | 2016-08-29,30,21,晴,西北风,2级
608 | 2016-08-30,30,21,晴,西北风,1级
609 | 2016-08-31,31,22,多云,西南风,2级
610 | 2016-09-01,33,23,多云,南风,2级
611 | 2016-09-02,34,22,晴,东风,1级
612 | 2016-09-03,34,23,晴,东北风,1级
613 | 2016-09-04,31,22,多云,东南风,2级
614 | 2016-09-05,32,23,多云,东风,1级
615 | 2016-09-06,27,21,阵雨,西南风,1级
616 | 2016-09-07,27,20,阵雨,东南风,2级
617 | 2016-09-08,30,21,多云,东南风,1级
618 | 2016-09-09,33,23,多云,东风,1级
619 | 2016-09-10,30,21,阴,东南风,2级
620 | 2016-09-11,28,20,多云,东南风,1级
621 | 2016-09-12,30,20,多云,东风,2级
622 | 2016-09-13,31,21,多云,东风,2级
623 | 2016-09-14,29,22,多云,东北风,1级
624 | 2016-09-15,24,21,大雨,东风,2级
625 | 2016-09-16,24,22,大到暴雨,北风,3级
626 | 2016-09-17,27,21,多云,东北风,2级
627 | 2016-09-18,28,20,多云,北风,3级
628 | 2016-09-19,26,18,多云,北风,2级
629 | 2016-09-20,25,17,多云,西北风,1级
630 | 2016-09-21,24,17,阴,东风,2级
631 | 2016-09-22,27,18,多云,东北风,2级
632 | 2016-09-23,27,19,多云,东风,2级
633 | 2016-09-24,28,20,多云,东风,2级
634 | 2016-09-25,28,21,多云,东风,2级
635 | 2016-09-26,29,22,多云,东风,2级
636 | 2016-09-27,30,23,阵雨,东北风,4级
637 | 2016-09-28,24,19,小到中雨,东北风,4级
638 | 2016-09-29,21,17,中雨,北风,3级
639 | 2016-09-30,22,19,中雨,东北风,3级
640 | 2016-10-01,25,21,小雨,东南风,微风
641 | 2016-10-02,25,20,阵雨,西风,2级
642 | 2016-10-03,27,18,多云,东北风,2级
643 | 2016-10-04,26,20,多云,北风,2级
644 | 2016-10-05,26,20,多云,东南风,1级
645 | 2016-10-06,25,19,多云,东风,2级
646 | 2016-10-07,20,16,中雨,北风,2级
647 | 2016-10-08,23,15,多云,东北风,1级
648 | 2016-10-09,21,13,多云,东北风,2级
649 | 2016-10-10,22,15,多云,东风,3级
650 | 2016-10-11,21,16,阴,东北风,2级
651 | 2016-10-12,21,16,阴,东风,1级
652 | 2016-10-13,20,16,阴,北风,2级
653 | 2016-10-14,19,17,阴,东风,2级
654 | 2016-10-15,20,16,阵雨,东风,1级
655 | 2016-10-16,23,17,阴,西北风,1级
656 | 2016-10-17,23,18,阴,东风,2级
657 | 2016-10-18,24,18,阴,东风,2级
658 | 2016-10-19,23,18,阴,东风,2级
659 | 2016-10-20,21,18,中雨,东风,2级
660 | 2016-10-21,22,19,中雨,东风,2级
661 | 2016-10-22,22,18,大雨,东北风,3级
662 | 2016-10-23,20,15,阵雨,东北风,2级
663 | 2016-10-24,21,15,阴,南风,1级
664 | 2016-10-25,19,16,阴,东风,1级
665 | 2016-10-26,20,16,大雨,东风,3级
666 | 2016-10-27,18,15,中雨,东南风,2级
667 | 2016-10-28,16,11,阵雨,北风,3级
668 | 2016-10-29,13,8,阵雨,西北风,2级
669 | 2016-10-30,18,12,阴,东风,1级
670 | 2016-10-31,15,10,阴,北风,4级
671 | 2016-11-01,15,5,多云,东风,2级
672 | 2016-11-02,16,8,晴,东南风,微风
673 | 2016-11-03,18,8,晴,南风,1级
674 | 2016-11-04,21,11,晴,东南风,2级
675 | 2016-11-05,23,11,晴,东南风,微风
676 | 2016-11-06,23,12,多云,东风,3级
677 | 2016-11-07,18,11,中雨,北风,5级
678 | 2016-11-08,12,8,阴,北风,2级
679 | 2016-11-09,12,8,阴,东北风,2级
680 | 2016-11-10,13,6,阴,西南风,1级
681 | 2016-11-11,18,8,多云,东南风,2级
682 | 2016-11-12,18,10,多云,北风,微风
683 | 2016-11-13,20,13,阴,东风,2级
684 | 2016-11-14,20,12,阴,东北风,3级
685 | 2016-11-15,17,7,多云,东风,2级
686 | 2016-11-16,17,11,阴,东风,1级
687 | 2016-11-17,16,14,阵雨,东风,1级
688 | 2016-11-18,19,15,小到中雨,东南风,1级
689 | 2016-11-19,19,15,阴,东北风,2级
690 | 2016-11-20,17,14,小雨,东风,2级
691 | 2016-11-21,17,10,小雨,东北风,2级
692 | 2016-11-22,9,2,小雨,北风,3级
693 | 2016-11-23,3,-2,中雪,西北风,2级
694 | 2016-11-24,5,0,多云,北风,1级
695 | 2016-11-25,8,3,阴,东北风,1级
696 | 2016-11-26,8,1,小雨,北风,微风
697 | 2016-11-27,13,2,晴,北风,微风
698 | 2016-11-28,13,5,多云,北风,2级
699 | 2016-11-29,12,7,多云,东风,2级
700 | 2016-11-30,13,4,阴,西南风,2级
701 | 2016-12-01,14,5,晴,东风,2级
702 | 2016-12-02,14,5,多云,东风,1级
703 | 2016-12-03,15,8,多云,北风,微风
704 | 2016-12-04,17,7,多云,西南风,1级
705 | 2016-12-05,19,5,多云,北风,4级
706 | 2016-12-06,10,1,多云,西风,1级
707 | 2016-12-07,15,5,多云,东南风,1级
708 | 2016-12-08,16,5,晴,西南风,1级
709 | 2016-12-09,13,5,晴,东北风,3级
710 | 2016-12-10,11,6,多云,东风,3级
711 | 2016-12-11,12,6,多云,东风,2级
712 | 2016-12-12,12,9,阵雨,西北风,1级
713 | 2016-12-13,11,3,小到中雨,北风,2级
714 | 2016-12-14,6,-2,多云,北风,3级
715 | 2016-12-15,8,-1,晴,西风,1级
716 | 2016-12-16,9,-1,晴,东南风,1级
717 | 2016-12-17,12,2,多云,东南风,2级
718 | 2016-12-18,14,5,多云,东南风,微风
719 | 2016-12-19,16,6,多云,东北风,2级
720 | 2016-12-20,12,9,小雨,东风,3级
721 | 2016-12-21,13,6,中雨,西北风,3级
722 | 2016-12-22,9,2,阴,北风,1级
723 | 2016-12-23,9,2,多云,北风,2级
724 | 2016-12-24,8,2,多云,东风,3级
725 | 2016-12-25,11,8,小到中雨,东风,2级
726 | 2016-12-26,10,1,小雨,西北风,3级
727 | 2016-12-27,5,-2,多云,北风,2级
728 | 2016-12-28,6,-1,晴,东北风,1级
729 | 2016-12-29,6,-1,多云,西北风,1级
730 | 2016-12-30,9,0,晴,东南风,1级
731 | 2016-12-31,11,-1,晴,北风,微风
732 | 


--------------------------------------------------------------------------------