├── .gitignore
├── .ipynb_checkpoints
    ├── A Look at Outliers-checkpoint.ipynb
    ├── Continuous Prior-checkpoint.ipynb
    ├── Logistic Gradient Descent-checkpoint.ipynb
    └── Python_For_Data_Mining-checkpoint.ipynb
├── Bayes_Primer.ipynb
├── Bayesian_Logistic_Regression.ipynb
├── Biggest_Misses.ipynb
├── Continuous Prior.ipynb
├── Get_Debate_Data.ipynb
├── Getting_Started_With_Spark_DataFrame.ipynb
├── LSTM_Debate_Model.ipynb
├── Logistic Gradient Descent.ipynb
├── NFL_Draft_Analyze.ipynb
├── NFL_Draft_Scrape.ipynb
├── Predict Fantasy Football Points.ipynb
├── Python_For_Data_Mining.ipynb
├── README.md
├── TalkPay.ipynb
├── Top100_OT.png
├── Top_Offensive_Tackles.ipynb
└── us_state_map_makefile


/.gitignore:
--------------------------------------------------------------------------------
1 | in_progress/
2 | data/
3 | .ipynb_checkpoints/
4 | 


--------------------------------------------------------------------------------
/Get_Debate_Data.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 |       "/Users/tylerfolkman/anaconda/lib/python2.7/site-packages/ipykernel/__main__.py:19: UnicodeWarning: Unicode equal comparison failed to convert both arguments to Unicode - interpreting them as being unequal\n"
 15 |      ]
 16 |     }
 17 |    ],
 18 |    "source": [
 19 |     "import requests\n",
 20 |     "from bs4 import BeautifulSoup\n",
 21 |     "from collections import defaultdict\n",
 22 |     "\n",
 23 |     "\n",
 24 |     "def text_to_dict(paragraph_array, d, candidates):\n",
 25 |     "    '''takes an array of text paragraphs from debate and returns dict \n",
 26 |     "    where key is person and value is list of text spoken by that candidate'''\n",
 27 |     "    # just a default speaker that won't end up in our returned data\n",
 28 |     "    # will get replaced when an actual speaker is found\n",
 29 |     "    speaker = \"<START>\"\n",
 30 |     "    for paragraph in paragraph_array:\n",
 31 |     "        words = paragraph.text.split(' ')\n",
 32 |     "        first_word = words[0]\n",
 33 |     "        # only new speaker when have SPEAKER: format\n",
 34 |     "        if first_word[-1] == \":\":\n",
 35 |     "            speaker = first_word[:-1]\n",
 36 |     "        # only keep candidates text\n",
 37 |     "        if speaker in candidates:\n",
 38 |     "            d[speaker].append(words[1:])\n",
 39 |     "    return d\n",
 40 |     "\n",
 41 |     "\n",
 42 |     "def process_url(url, speaker_dict, candidates):\n",
 43 |     "    # requests gets the source code from the url and extracts it as text\n",
 44 |     "    html = requests.get(url).text\n",
 45 |     "    # beautifulsoup is a library that takes in text source code and returns a structured format of that\n",
 46 |     "    # source code that you can more easily search and parse.\n",
 47 |     "    soup = BeautifulSoup(html, 'html5lib')\n",
 48 |     "    # get all the 'p' tags from the source with class = 'story-body-text'\n",
 49 |     "    # this was determined by looking at the source code\n",
 50 |     "    # the first and last paragraphs are intro and ending\n",
 51 |     "    paragraphs = soup('p', {'class': 'story-body-text'})[1:-1]\n",
 52 |     "    text_to_dict(paragraphs, speaker_dict, candidates)\n",
 53 |     "    \n",
 54 |     "    \n",
 55 |     "def process_url_list(urls, speaker_dict, candidates):\n",
 56 |     "    for url in urls:\n",
 57 |     "        process_url(url, speaker_dict, candidates)\n",
 58 |     "    \n",
 59 |     "    \n",
 60 |     "candidates = ['BUSH', 'TRUMP', 'RUBIO', 'CARSON', 'FIORINA', 'KASICH', 'CRUZ', 'PAUL',\n",
 61 |     "             'SANDERS', 'CLINTON', \"O’MALLEY\"]\n",
 62 |     "urls = ['http://www.nytimes.com/2015/11/11/us/politics/transcript-republican-presidential-debate.html',\n",
 63 |     "       'http://www.nytimes.com/2016/02/14/us/politics/transcript-of-the-republican-presidential-debate.html',\n",
 64 |     "       'http://www.nytimes.com/2016/01/15/us/politics/transcript-of-republican-presidential-debate.html',\n",
 65 |     "       'http://www.nytimes.com/2016/01/29/us/politics/republican-presidential-debate-transcript.html?_r=0',\n",
 66 |     "       'http://www.nytimes.com/2016/01/18/us/politics/transcript-of-the-democratic-presidential-debate.html',\n",
 67 |     "       'http://www.nytimes.com/2016/02/12/us/politics/transcript-of-the-democratic-presidential-debate-in-milwaukee.html',\n",
 68 |     "       'http://www.nytimes.com/2016/02/07/us/politics/transcript-of-the-republican-presidential-debate-in-new-hampshire.html',\n",
 69 |     "       'http://www.nytimes.com/2016/02/05/us/politics/transcript-of-the-democratic-presidential-debate.html',\n",
 70 |     "       'http://www.nytimes.com/2016/01/29/us/politics/republican-presidential-debate-preliminary-transcript.html',\n",
 71 |     "       'http://www.nytimes.com/2015/12/16/us/politics/transcript-main-republican-presidential-debate.html']\n",
 72 |     "speaker_dict = defaultdict(list)\n",
 73 |     "\n",
 74 |     "process_url_list(urls, speaker_dict, candidates)"
 75 |    ]
 76 |   },
 77 |   {
 78 |    "cell_type": "code",
 79 |    "execution_count": 2,
 80 |    "metadata": {
 81 |     "collapsed": false
 82 |    },
 83 |    "outputs": [],
 84 |    "source": [
 85 |     "def pargraphs_to_words(speaker_dict):\n",
 86 |     "    d = defaultdict(list)\n",
 87 |     "    for candidate, paragraphs in speaker_dict.items():\n",
 88 |     "        for paragraph in paragraphs:\n",
 89 |     "            for word in paragraph:\n",
 90 |     "                d[candidate].append(word)\n",
 91 |     "    return d"
 92 |    ]
 93 |   },
 94 |   {
 95 |    "cell_type": "code",
 96 |    "execution_count": 3,
 97 |    "metadata": {
 98 |     "collapsed": false
 99 |    },
100 |    "outputs": [],
101 |    "source": [
102 |     "candidate_words = pargraphs_to_words(speaker_dict)"
103 |    ]
104 |   },
105 |   {
106 |    "cell_type": "code",
107 |    "execution_count": 4,
108 |    "metadata": {
109 |     "collapsed": true
110 |    },
111 |    "outputs": [],
112 |    "source": [
113 |     "import json\n",
114 |     "\n",
115 |     "with open('candidate_words_dict.json', 'w') as f:\n",
116 |     "    json.dump(candidate_words, f)"
117 |    ]
118 |   }
119 |  ],
120 |  "metadata": {
121 |   "kernelspec": {
122 |    "display_name": "Python 2",
123 |    "language": "python",
124 |    "name": "python2"
125 |   },
126 |   "language_info": {
127 |    "codemirror_mode": {
128 |     "name": "ipython",
129 |     "version": 2
130 |    },
131 |    "file_extension": ".py",
132 |    "mimetype": "text/x-python",
133 |    "name": "python",
134 |    "nbconvert_exporter": "python",
135 |    "pygments_lexer": "ipython2",
136 |    "version": "2.7.11"
137 |   }
138 |  },
139 |  "nbformat": 4,
140 |  "nbformat_minor": 0
141 | }
142 | 


--------------------------------------------------------------------------------
/Getting_Started_With_Spark_DataFrame.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "#Spark\n",
  8 |     "\n",
  9 |     "[Spark](https://spark.apache.org/) is a really awesome tool to easily do distributed computations in order to process large-scale data. To be honest, most people probably don't need spark for their own side projects - most of these data will fit in memory or work well in a traditional database like PostgreSQL. That being said, there is a good chance you might need Spark if you are doing data science type work for your job. A lot of companies have a tremendous amount of data and Spark is a great tool to help effectively process these large data.\n",
 10 |     "\n",
 11 |     "In case you are not familiar with the map reduce structure, here is a very [brief introduction](http://hci.stanford.edu/courses/cs448g/a2/files/map_reduce_tutorial.pdf). Spark is based on this map reduce paradigm, but has made some nice improvements to the open source version Hadoop. A few of these improvements include the ability to cache data to memory, a simpler API supported in multiple languages (scala, python, and java I believe), and some really nice libraries - including a machine learning and SQL library. In my opinion, these additions really make Spark a powerful tool with a realtively easy learning curve.\n",
 12 |     "\n",
 13 |     "My goal today is to show you how to get started with Spark and get introduced to Spark data frames."
 14 |    ]
 15 |   },
 16 |   {
 17 |    "cell_type": "markdown",
 18 |    "metadata": {},
 19 |    "source": [
 20 |     "#Installation\n",
 21 |     "\n",
 22 |     "Fortunately, you don't need a cluster to get started with Spark. You can run it locally as well. This obviously doesn't provide the benefits of a cluster, but is great for learning.\n",
 23 |     "\n",
 24 |     "Go ahead and install the latest verison of Spark (I would recommend pre-built with latest version of Hadoop) [here](https://spark.apache.org/downloads.html).\n",
 25 |     "\n",
 26 |     "Once you download and unpack Spark, you can navigate to that directory. From there you can go to bin/pyspark to launch an interactive version of pyspark. This is really useful for experimentation, but let's see if we can't get Spark to work in your ipython notebook.\n",
 27 |     "\n",
 28 |     "Follow the \"Create PySpark Profile for IPython\" [here](http://ramhiser.com/2015/02/01/configuring-ipython-notebook-support-for-pyspark/) **with the exception of changing PYSPARK_SUBMIT_ARGS to \"--master local[2] pyspark-shell\"** Adding the pyspark-shell to the end is now necessary.\n",
 29 |     "\n",
 30 |     "You should now be able to run: ipython notebook --profile=pyspark. To test that it works make sure you can do the following:"
 31 |    ]
 32 |   },
 33 |   {
 34 |    "cell_type": "code",
 35 |    "execution_count": 11,
 36 |    "metadata": {
 37 |     "collapsed": false
 38 |    },
 39 |    "outputs": [
 40 |     {
 41 |      "data": {
 42 |       "text/plain": [
 43 |        "<pyspark.context.SparkContext at 0x1072eebd0>"
 44 |       ]
 45 |      },
 46 |      "execution_count": 11,
 47 |      "metadata": {},
 48 |      "output_type": "execute_result"
 49 |     }
 50 |    ],
 51 |    "source": [
 52 |     "sc"
 53 |    ]
 54 |   },
 55 |   {
 56 |    "cell_type": "code",
 57 |    "execution_count": 12,
 58 |    "metadata": {
 59 |     "collapsed": true
 60 |    },
 61 |    "outputs": [],
 62 |    "source": [
 63 |     "a=sc.parallelize([1,2,3,4])"
 64 |    ]
 65 |   },
 66 |   {
 67 |    "cell_type": "code",
 68 |    "execution_count": 13,
 69 |    "metadata": {
 70 |     "collapsed": false
 71 |    },
 72 |    "outputs": [
 73 |     {
 74 |      "data": {
 75 |       "text/plain": [
 76 |        "[1, 2, 3, 4]"
 77 |       ]
 78 |      },
 79 |      "execution_count": 13,
 80 |      "metadata": {},
 81 |      "output_type": "execute_result"
 82 |     }
 83 |    ],
 84 |    "source": [
 85 |     "a.collect()"
 86 |    ]
 87 |   },
 88 |   {
 89 |    "cell_type": "markdown",
 90 |    "metadata": {},
 91 |    "source": [
 92 |     "You should also now be able to go to http://localhost:4040/ and see the spark application UI. Hopefully you are now up and running with Spark in your IPython notebook! Also, as a quick side note, if you prefer using an IDE, check out this [post](https://gist.github.com/bigaidream/40fe0f8267a80e7c9cf8) on how to get autocompletion for Spark in PyCharm."
 93 |    ]
 94 |   },
 95 |   {
 96 |    "cell_type": "markdown",
 97 |    "metadata": {
 98 |     "collapsed": true
 99 |    },
100 |    "source": [
101 |     "#[NY Data](https://data.ny.gov/)\n",
102 |     "New York provides a lot of open data that is easily accessible. A great resource to discovery some really interesting things about NY as well as cities in general. To showcase some of the Spark features, I will be taking a look at salary information for state authorities.\n",
103 |     "\n",
104 |     "We are first going to use their API to grab up to 50,000 observations, convert the data to JSON and save to disk. We can then load the file into a Spark data frame."
105 |    ]
106 |   },
107 |   {
108 |    "cell_type": "code",
109 |    "execution_count": 14,
110 |    "metadata": {
111 |     "collapsed": true
112 |    },
113 |    "outputs": [],
114 |    "source": [
115 |     "import urllib2\n",
116 |     "import json\n",
117 |     "req = urllib2.Request('https://data.ny.gov/resource/unag-2p27.json?$limit=50000')\n",
118 |     "resp = urllib2.urlopen(req)\n",
119 |     "content = json.loads(resp.read())\n",
120 |     "with open(\"data/ny_salaries.json\", \"w\") as f:\n",
121 |     "    json.dump(content, f)"
122 |    ]
123 |   },
124 |   {
125 |    "cell_type": "markdown",
126 |    "metadata": {},
127 |    "source": [
128 |     "#[Spark Data Frames](http://spark.apache.org/docs/latest/api/python/pyspark.sql.html)"
129 |    ]
130 |   },
131 |   {
132 |    "cell_type": "code",
133 |    "execution_count": 15,
134 |    "metadata": {
135 |     "collapsed": false
136 |    },
137 |    "outputs": [],
138 |    "source": [
139 |     "from pyspark.sql import SQLContext\n",
140 |     "sqlContext = SQLContext(sc)\n",
141 |     "df = sqlContext.read.json(\"data/ny_salaries.json\")"
142 |    ]
143 |   },
144 |   {
145 |    "cell_type": "markdown",
146 |    "metadata": {},
147 |    "source": [
148 |     "Lets take a look at the schema"
149 |    ]
150 |   },
151 |   {
152 |    "cell_type": "code",
153 |    "execution_count": 16,
154 |    "metadata": {
155 |     "collapsed": false
156 |    },
157 |    "outputs": [
158 |     {
159 |      "name": "stdout",
160 |      "output_type": "stream",
161 |      "text": [
162 |       "root\n",
163 |       " |-- actual_salary_paid: string (nullable = true)\n",
164 |       " |-- authority_name: string (nullable = true)\n",
165 |       " |-- base_annualized_salary: string (nullable = true)\n",
166 |       " |-- department: string (nullable = true)\n",
167 |       " |-- exempt_indicator: string (nullable = true)\n",
168 |       " |-- extra_pay: string (nullable = true)\n",
169 |       " |-- first_name: string (nullable = true)\n",
170 |       " |-- fiscal_year_end_date: string (nullable = true)\n",
171 |       " |-- group: string (nullable = true)\n",
172 |       " |-- has_employees: string (nullable = true)\n",
173 |       " |-- last_name: string (nullable = true)\n",
174 |       " |-- middle_initial: string (nullable = true)\n",
175 |       " |-- other_compensation: string (nullable = true)\n",
176 |       " |-- overtime_paid: string (nullable = true)\n",
177 |       " |-- paid_by_another_entity: string (nullable = true)\n",
178 |       " |-- paid_by_state_or_local_government: string (nullable = true)\n",
179 |       " |-- pay_type: string (nullable = true)\n",
180 |       " |-- performance_bonus: string (nullable = true)\n",
181 |       " |-- title: string (nullable = true)\n",
182 |       " |-- total_compensation: string (nullable = true)\n",
183 |       "\n"
184 |      ]
185 |     }
186 |    ],
187 |    "source": [
188 |     "df.printSchema()"
189 |    ]
190 |   },
191 |   {
192 |    "cell_type": "markdown",
193 |    "metadata": {},
194 |    "source": [
195 |     "And now a specific column"
196 |    ]
197 |   },
198 |   {
199 |    "cell_type": "code",
200 |    "execution_count": 17,
201 |    "metadata": {
202 |     "collapsed": false
203 |    },
204 |    "outputs": [
205 |     {
206 |      "name": "stdout",
207 |      "output_type": "stream",
208 |      "text": [
209 |       "+------------------+\n",
210 |       "|total_compensation|\n",
211 |       "+------------------+\n",
212 |       "|          28725.55|\n",
213 |       "|          31973.09|\n",
214 |       "|          68654.00|\n",
215 |       "|          25826.95|\n",
216 |       "|          60114.08|\n",
217 |       "|          15257.43|\n",
218 |       "|          85000.00|\n",
219 |       "|          44454.64|\n",
220 |       "|          39474.83|\n",
221 |       "|           2240.00|\n",
222 |       "|          56267.98|\n",
223 |       "|           1817.60|\n",
224 |       "|           4530.33|\n",
225 |       "|          59186.44|\n",
226 |       "|          57664.04|\n",
227 |       "|           1333.50|\n",
228 |       "|          63339.92|\n",
229 |       "|          59782.21|\n",
230 |       "|          48451.11|\n",
231 |       "|          68632.35|\n",
232 |       "+------------------+\n",
233 |       "\n"
234 |      ]
235 |     }
236 |    ],
237 |    "source": [
238 |     "df.select(\"total_compensation\").show()"
239 |    ]
240 |   },
241 |   {
242 |    "cell_type": "markdown",
243 |    "metadata": {},
244 |    "source": [
245 |     "We can easily filter by column"
246 |    ]
247 |   },
248 |   {
249 |    "cell_type": "code",
250 |    "execution_count": 18,
251 |    "metadata": {
252 |     "collapsed": false
253 |    },
254 |    "outputs": [
255 |     {
256 |      "data": {
257 |       "text/plain": [
258 |        "Row(actual_salary_paid=u'133040.96', authority_name=u'Battery Park City Authority', base_annualized_salary=u'133041.00', department=u'Administration/BPCPC', exempt_indicator=u'Y', extra_pay=u'0.00', first_name=u'Tessa', fiscal_year_end_date=u'2011-10-31T00:00:00', group=u'Executive', has_employees=None, last_name=u'Huxley', middle_initial=None, other_compensation=u'750.00', overtime_paid=u'0.00', paid_by_another_entity=u'N', paid_by_state_or_local_government=None, pay_type=u'FT', performance_bonus=u'0.00', title=u'Executive Director', total_compensation=u'133790.96')"
259 |       ]
260 |      },
261 |      "execution_count": 18,
262 |      "metadata": {},
263 |      "output_type": "execute_result"
264 |     }
265 |    ],
266 |    "source": [
267 |     "df.filter(df[\"total_compensation\"] > 100000).first()"
268 |    ]
269 |   },
270 |   {
271 |    "cell_type": "markdown",
272 |    "metadata": {},
273 |    "source": [
274 |     "Now lets take a look at the highest paying departments on average. We first make sure to convert the column to float, and then use Spark's groupby and orderby operations to get the data we need."
275 |    ]
276 |   },
277 |   {
278 |    "cell_type": "code",
279 |    "execution_count": 19,
280 |    "metadata": {
281 |     "collapsed": false
282 |    },
283 |    "outputs": [],
284 |    "source": [
285 |     "from pyspark.sql.functions import desc\n",
286 |     "df_cast = df.withColumn(\"total_compensation_float\", df.total_compensation.astype(\"float\"))\n",
287 |     "top10 = (df_cast\n",
288 |     "         .groupBy(\"department\")\n",
289 |     "         .mean(\"total_compensation_float\")\n",
290 |     "         .orderBy(desc(\"AVG(total_compensation_float)\"))\n",
291 |     "         .take(10))"
292 |    ]
293 |   },
294 |   {
295 |    "cell_type": "code",
296 |    "execution_count": 20,
297 |    "metadata": {
298 |     "collapsed": false
299 |    },
300 |    "outputs": [
301 |     {
302 |      "data": {
303 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAGcCAYAAAA2+rwbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nNIWIiBhLU4iIiLE0hYiIGEtTiIiIsf8AZqxFK4vyNDsAAAAASUVORK5CYII=\n",
304 |       "text/plain": [
305 |        "<matplotlib.figure.Figure at 0x10d592490>"
306 |       ]
307 |      },
308 |      "metadata": {},
309 |      "output_type": "display_data"
310 |     }
311 |    ],
312 |    "source": [
313 |     "import seaborn as sns\n",
314 |     "import numpy as np\n",
315 |     "import matplotlib.pyplot as plt\n",
316 |     "%matplotlib inline\n",
317 |     "\n",
318 |     "department = np.empty(10, dtype=\"S30\")\n",
319 |     "salary = np.empty(10)\n",
320 |     "for i, row in enumerate(top10):\n",
321 |     "    department[i] = row[0]\n",
322 |     "    salary[i] = row[1]\n",
323 |     "sns.barplot(x=department, y=salary, x_order=department)\n",
324 |     "plt.xticks(rotation='vertical')\n",
325 |     "sns.despine()"
326 |    ]
327 |   },
328 |   {
329 |    "cell_type": "markdown",
330 |    "metadata": {},
331 |    "source": [
332 |     "Looks like a nice place to be is the Office of Strategic Initative...but I wonder how many people are in that department in our data? Let's utilize the SQL query method on our data frame\n",
333 |     "#To SQL"
334 |    ]
335 |   },
336 |   {
337 |    "cell_type": "code",
338 |    "execution_count": 21,
339 |    "metadata": {
340 |     "collapsed": false
341 |    },
342 |    "outputs": [
343 |     {
344 |      "data": {
345 |       "text/plain": [
346 |        "[Row(c0=1)]"
347 |       ]
348 |      },
349 |      "execution_count": 21,
350 |      "metadata": {},
351 |      "output_type": "execute_result"
352 |     }
353 |    ],
354 |    "source": [
355 |     "df_cast.registerTempTable(\"df_cast\")\n",
356 |     "sqlContext.sql(\"select count(*) from df_cast where department = 'Office of Strategic Initiative'\").collect()"
357 |    ]
358 |   },
359 |   {
360 |    "cell_type": "markdown",
361 |    "metadata": {},
362 |    "source": [
363 |     "Looks like there is only one lucky person :). We could also convert our Spark data frame to a pandas data frame and do the same calculation. Nice.\n",
364 |     "#To Pandas"
365 |    ]
366 |   },
367 |   {
368 |    "cell_type": "code",
369 |    "execution_count": 22,
370 |    "metadata": {
371 |     "collapsed": false
372 |    },
373 |    "outputs": [
374 |     {
375 |      "data": {
376 |       "text/plain": [
377 |        "1"
378 |       ]
379 |      },
380 |      "execution_count": 22,
381 |      "metadata": {},
382 |      "output_type": "execute_result"
383 |     }
384 |    ],
385 |    "source": [
386 |     "df_pandas = df_cast.toPandas()\n",
387 |     "df_pandas.department.value_counts()['Office of Strategic Initiative']"
388 |    ]
389 |   },
390 |   {
391 |    "cell_type": "markdown",
392 |    "metadata": {},
393 |    "source": [
394 |     "#Summary Statistics / Functions"
395 |    ]
396 |   },
397 |   {
398 |    "cell_type": "code",
399 |    "execution_count": 23,
400 |    "metadata": {
401 |     "collapsed": false
402 |    },
403 |    "outputs": [
404 |     {
405 |      "name": "stdout",
406 |      "output_type": "stream",
407 |      "text": [
408 |       "+-------+------------------------+\n",
409 |       "|summary|total_compensation_float|\n",
410 |       "+-------+------------------------+\n",
411 |       "|  count|                   49997|\n",
412 |       "|   mean|       63374.58037148814|\n",
413 |       "| stddev|       35402.40384989916|\n",
414 |       "|    min|                     0.0|\n",
415 |       "|    max|               735124.75|\n",
416 |       "+-------+------------------------+\n",
417 |       "\n"
418 |      ]
419 |     }
420 |    ],
421 |    "source": [
422 |     "df_cast.describe('total_compensation_float').show()"
423 |    ]
424 |   },
425 |   {
426 |    "cell_type": "markdown",
427 |    "metadata": {},
428 |    "source": [
429 |     "Wow - that is a high max. Lets take a look at the departments by max salary. Excuse the pandas oneliner - would just be the same Spark code as used above to get the average"
430 |    ]
431 |   },
432 |   {
433 |    "cell_type": "code",
434 |    "execution_count": 24,
435 |    "metadata": {
436 |     "collapsed": false
437 |    },
438 |    "outputs": [
439 |     {
440 |      "data": {
441 |       "text/plain": [
442 |        "<matplotlib.axes._subplots.AxesSubplot at 0x10ca318d0>"
443 |       ]
444 |      },
445 |      "execution_count": 24,
446 |      "metadata": {},
447 |      "output_type": "execute_result"
448 |     },
449 |     {
450 |      "data": {
451 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAGrCAYAAAAvhYsOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3XmcXFWZ//FPFsJiEmQJAhpxwuh3UAcRRZZgQIdFGAUd\nRwTEQUZBICIouAUcFkF2BQTZIgKCiqC4ISQDPyExsgRkUAS/7BgFJZCGJLIESP/+OKeqK01vkK5z\nL9Tzfr3yStXt23WeSrrrufcszxnR3d1NCCGEADCy6gBCCCHURySFEEIITZEUQgghNEVSCCGE0BRJ\nIYQQQlMkhRBCCE2jB/qipJHAdOBNwFJgb+B54Pz8/HZgqu1uSXsD+wDPAUfbvkLSysBFwARgEbCn\n7UclbQacks+dafuo3N7hwI75+EG25w7z+w0hhDCAwe4UtgNeZXtL4Cjg68DJwDTbU4ARwM6S1gYO\nALYAtgeOlTQG2A+4LZ97IXBYft2zgN3y624qaSNJGwNTbG8K7AqcMZxvNIQQwuAGSwpPAatKGgGs\nCiwB3mF7Vv76lcA2wCbAHNvP2l4I3ANsCEwGrsrnXgVsI2kcMMb2/fn4jPwak4GZALbnAaMlrTEM\n7zGEEMIQDdh9BMwBVgL+BKwBfACY0vL1RaRkMR54op/jCwc41jg+CXgaeKyP12g9FkIIoY0GSwpf\nJN0BHCrpdcCvgRVavj4eeJz0IT+u5fi4Po73daz1NZb08xr9eu6557tHjx41yFsIIYTQy4j+vjBY\nUngVPVf1Xfn8WyVtZfs6YAfgGuAm4BhJK5LuLDYgDULPIQ0cz83nzrK9SNISSZOA+0njFkeQBrBP\nkHQSMBEYaXvBQMF1dT05SPiDmzBhHPPnL1ru13m5x1CXOOoQQ13iqEMMdYmjDjHUJY7hiGHChHH9\nfm2wpHAi8F1Js0l3CF8BbgHOzQPJdwCX5dlHpwGzSeMU02w/I+lM4IL8/c8Au+fX3Re4GBgFzGjM\nMsrnXZ9fY/+X8mZDCCG8dCNezlVS589ftNzBv1Iy/ysljjrEUJc46hBDXeKoQwx1iWOY7hT67T6K\nxWshhBCaIimEEEJoiqQQQgihKZJCCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQgghhKZICiGEEJoi\nKYQQQmiKpBBCCKEpkkIIIYSmSAohhBCaBiud/bK2ZMkS5s17cMBzurrGsmDB4n6/PnHieowZM2a4\nQwshhFp6RSeFefMe5MATf84qq671kr7/ySce4dQv7MT6679xmCMLIYR6ekUnBYBVVl2Lsau9tuow\nQgjhZeEVnxSqNhxdWBDdWCGEMiIptNnydmFBdGOFEMqJpFBAdGGFEF4uYkpqCCGEpkHvFCTtCXwi\nP10ZeBuwJXAqsBS4HZhqu1vS3sA+wHPA0bavkLQycBEwAVgE7Gn7UUmbAafkc2faPiq3dziwYz5+\nkO25w/VmQwghDGzQOwXbF9h+j+33ADcDBwD/A0yzPQUYAewsae38tS2A7YFjJY0B9gNuy+deCByW\nX/osYDfbWwKbStpI0sbAFNubArsCZwznmw0hhDCwIY8pSHon8Gbbn5F0hO1Z+UtXAtsBzwNzbD8L\nPCvpHmBDYDJwfD73KuCrksYBY2zfn4/PALYBngFmAtieJ2m0pDVsP7Z8bzPEQr4QwlC8mIHmacCR\n+fGIluOLgFWB8cAT/RxfOMCxxvFJwNPAY328RiSF5RQL+UIIQzGkpCDp1cCbbF+XDy1t+fJ44HHS\nh/y4luPj+jje17HW11jSz2v0abXVVmH06FH9xt3VNbbfrw3V6quPZcKEcYOf2MYYhiuO5Z0Ftbwx\nDFWJNoaiDnHUIQaoRxx1iAHqEUc7YxjqncIU4JqW57dK2ioniR3y124CjpG0IrASsAFpEHoOaeB4\nbj53lu1FkpZImgTcT+p+OoLUBXWCpJOAicBI2wv6C6qr68kBgx5sQdhQLFiwmPnzFy3X9w+HOsSx\nvDEMxYQJ49rexssljjrEUJc46hBDXeIYjhgGSipDTQpvAu5teX4wcG4eSL4DuCzPPjoNmE0awJ5m\n+xlJZwIXSJpNGjPYPb/GvsDFwChgRmOWUT7v+vwa+w8xvhBCCMNgSEnB9km9nt8NbN3HedOB6b2O\nPQXs0se5NwKb93H8SHrGLkIIIRQUi9dCCCE0RVIIIYTQFEkhhBBCUySFEEIITZEUQgghNEVSCCGE\n0BRJIYQQQlMkhRBCCE2RFEIIITTFdpyhmOEo3w1RwjuEdoqkEIpZ3vLdECW8Q2i3SAqhqOUt3x1C\naK8YUwghhNAUSSGEEEJTJIUQQghNkRRCCCE0RVIIIYTQFEkhhBBCUySFEEIITZEUQgghNA26eE3S\nV4APACsApwNzgPOBpcDtwFTb3ZL2BvYBngOOtn2FpJWBi4AJwCJgT9uPStoMOCWfO9P2Ubmtw4Ed\n8/GDbM8dzjcbQghhYAPeKUjaGtjc9hbA1sAk4GRgmu0pwAhgZ0lrAwcAWwDbA8dKGgPsB9yWz70Q\nOCy/9FnAbra3BDaVtJGkjYEptjcFdgXOGNZ3GkIIYVCDdR9tB/xB0k+BXwA/B95he1b++pXANsAm\nwBzbz9peCNwDbAhMBq7K514FbCNpHDDG9v35+Iz8GpOBmQC25wGjJa0xDO8xhBDCEA3WfTQBmAi8\nn3SX8AvS3UHDImBVYDzwRD/HFw5wrHF8EvA08Fgfr9F6LIQQQhsNlhQeBe60/Rxwl6SngdZqZuOB\nx0kf8uNajo/r43hfx1pfY0k/r9Gv1VZbhdGjR/X79a6usQN9+5CsvvpYJkwYN/iJbYyhLnHUIYbh\niGOoSrTxcogB6hFHHWKAesTRzhgGSwq/AQ4EviFpXWAV4BpJW9m+DtgBuAa4CThG0orASsAGpEHo\nOaSB47n53Fm2F0laImkScD+pi+oI4HngBEknke5ORtpeMFBwXV1PDhj8YHX5h2LBgsXMn79oub5/\nONQhjjrEMBxxDMWECePa3sbLIYa6xFGHGOoSx3DEMFBSGTAp5BlEUyTdRBp/2B94ADg3DyTfAVyW\nZx+dBszO502z/YykM4ELJM0GngF2zy+9L3AxMAqY0ZhllM+7vqWtEEIIBQ06JdX2l/o4vHUf500H\npvc69hSwSx/n3ghs3sfxI4EjB4sphBBCe8TitRBCCE2RFEIIITRFUgghhNAUSSGEEEJTJIUQQghN\nkRRCCCE0RVIIIYTQFEkhhBBCUySFEEIITZEUQgghNEVSCCGE0BRJIYQQQlMkhRBCCE2RFEIIITRF\nUgghhNAUSSGEEEJTJIUQQghNkRRCCCE0RVIIIYTQFEkhhBBC0+ihnCTpd8AT+el9wLHA+cBS4HZg\nqu1uSXsD+wDPAUfbvkLSysBFwARgEbCn7UclbQacks+dafuo3NbhwI75+EG25w7LOw0hhDCoQe8U\nJK0EYPs9+c8ngW8A02xPAUYAO0taGzgA2ALYHjhW0hhgP+C2fO6FwGH5pc8CdrO9JbCppI0kbQxM\nsb0psCtwxnC+2RBCCAMbSvfR24BVJM2QdE2+wt/Y9qz89SuBbYBNgDm2n7W9ELgH2BCYDFyVz70K\n2EbSOGCM7fvz8Rn5NSYDMwFszwNGS1pjud9lCCGEIRlKUvgHcKLt7YF9gYt7fX0RsCownp4upt7H\nFw5wbCivEUIIoYChjCncRbrqx/bdkh4D3t7y9fHA46QP+XEtx8f1cbyvY62vsaSf1+jTaqutwujR\no/oNvKtr7ABva2hWX30sEyaMG/zENsZQlzjqEMNwxDFUJdp4OcQA9YijDjFAPeJoZwxDSQp7kbqB\npkpal/RBPVPSVravA3YArgFuAo6RtCKwErABaRB6DmngeG4+d5btRZKWSJoE3A9sBxwBPA+cIOkk\nYCIw0vaC/gLr6npywMAXLFg8hLc3sAULFjN//qLl+v7hUIc46hDDcMQxFBMmjGt7Gy+HGOoSRx1i\nqEscwxHDQEllKEnhO8B3JTXGEPYCHgPOzQPJdwCX5dlHpwGzSd1S02w/I+lM4AJJs4FngN3z6zS6\nokYBMxqzjPJ51+fX2P9FvdMQQgjLZdCkYPs54ON9fGnrPs6dDkzvdewpYJc+zr0R2LyP40cCRw4W\nVwghhOEXi9dCCCE0RVIIIYTQFEkhhBBC05DKXITwSrJkyRLmzXtwwHO6usYOOFtq4sT1GDNmzHCH\nFkLlIimEjjNv3oMceOLPWWXVtV7S9z/5xCOc+oWdWH/9Nw5zZCFUL5JC6EirrLoWY1d7bdVhhFA7\nMaYQQgihKZJCCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQgghhKZICiGEEJoiKYQQQmiKpBBCCKEp\nkkIIIYSmSAohhBCaIimEEEJoiqQQQgihKZJCCCGEpiGVzpa0FnAL8G/AUuD8/PftwFTb3ZL2BvYB\nngOOtn2FpJWBi4AJwCJgT9uPStoMOCWfO9P2Ubmdw4Ed8/GDbM8dtncaQghhUIPeKUhaATgb+Acw\nAvgGMM32lPx8Z0lrAwcAWwDbA8dKGgPsB9yWz70QOCy/7FnAbra3BDaVtJGkjYEptjcFdgXOGMb3\nGUIIYQiG0n10InAm8HB+vrHtWfnxlcA2wCbAHNvP2l4I3ANsCEwGrsrnXgVsI2kcMMb2/fn4jPwa\nk4GZALbnAaMlrbE8by6EEMKLM2BSkPQJYL7tmfnQiPynYRGwKjAeeKKf4wsHODaU1wghhFDIYGMK\newHdkrYBNgIuII0PNIwHHid9yI9rOT6uj+N9HWt9jSX9vEa/VlttFUaPHtXv17u6xg707UOy+upj\nmTBh3OAntjGGusRRhxjqEsfyxjBUJdoYijrEUYcYoB5xtDOGAZOC7a0ajyX9GtgXOFHSVravA3YA\nrgFuAo6RtCKwErABaRB6DmngeG4+d5btRZKWSJoE3A9sBxwBPA+cIOkkYCIw0vaCgeLr6npywDe3\nYMHiAb8+FAsWLGb+/EXL9f3DoQ5x1CGGusSxvDEMxYQJ49rexssljjrEUJc4hiOGgZLKkGYftegG\nDgbOzQPJdwCX5dlHpwGzSV1S02w/I+lM4AJJs4FngN3z6+wLXAyMAmY0Zhnl867Pr7H/i4wthBDC\nchpyUrD9npanW/fx9enA9F7HngJ26ePcG4HN+zh+JHDkUGMK4eVqyZIlzJv34IDndHWNHfSuZuLE\n9RgzZsxwhhY63Iu9UwghDIN58x7kwBN/ziqrrvWSX+PJJx7h1C/sxPrrv3EYIwudLpJCCBVZZdW1\nGLvaa6sOI4RlRFIIoUMNpQsLBu/Gii6sV5ZICiF0qOjCCn2JpBBCB6tDF1YMutdLJIUQQqXijqVe\nIimEECpXhzuWkMR+CiGEEJoiKYQQQmiKpBBCCKEpkkIIIYSmSAohhBCaIimEEEJoiqQQQgihKZJC\nCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQgghhKZICiGEEJoGrZIqaRRwLvAmoBvYF3gGOB9YCtwO\nTLXdLWlvYB/gOeBo21dIWhm4CJgALAL2tP2opM2AU/K5M20flds7HNgxHz/I9txhfL8hhBAGMJQ7\nhfcDS21vCRwGfB04GZhmewowAthZ0trAAcAWwPbAsZLGAPsBt+VzL8yvAXAWsFt+3U0lbSRpY2CK\n7U2BXYEzhuuNhhBCGNygdwq2fybpl/npG4AuYBvbs/KxK4HtgOeBObafBZ6VdA+wITAZOD6fexXw\nVUnjgDG278/HZwDbkO5AZuZ250kaLWkN248t5/sMIYR+1WW/6jrsQjekTXZsPy/pfOCDwEeAbVu+\nvAhYFRgPPNHP8YUDHGscnwQ8DTzWx2tEUgghtE1ddn+rQxxD3nnN9ickvQa4CVip5UvjgcdJH/Lj\nWo6P6+N4X8daX2NJP6/Rp9VWW4XRo0f1G3NX19gB39NQrL76WCZMGDf4iW2MoS5x1CGGusRRhxiW\nN446xFCXOLq6xg7L7m/D8W9RdRxDGWj+OPA628cCT5G6iW6WtJXt64AdgGtIyeIYSSuSksYGpEHo\nOaSB47n53Fm2F0laImkScD+p++mI/NonSDoJmAiMtL2gv9i6up4cMPbBbrGGYsGCxcyfv2i5vn84\n1CGOOsRQlzjqEMPyxlGHGOoSRx1iKBnHQAljKHcKlwHnS7oOWAE4EPgTcG4eSL4DuCzPPjoNmE0a\nwJ5m+xlJZwIXSJpNGjPYPb/uvsDFwChgRmOWUT7v+vwa+w8hvhBCCMNkKAPNTwEf7eNLW/dx7nRg\neh/fv0sf594IbN7H8SOBIweLK4QQwvCLxWshhBCaIimEEEJoiqQQQgihKZJCCCGEpkgKIYQQmiIp\nhBBCaIqkEEIIoSmSQgghhKZICiGEEJoiKYQQQmiKpBBCCKEpkkIIIYSmSAohhBCaIimEEEJoiqQQ\nQgihKZJCCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQgghhKbRA31R0grAecB6wIrA0cCdwPnAUuB2\nYKrtbkl7A/sAzwFH275C0srARcAEYBGwp+1HJW0GnJLPnWn7qNze4cCO+fhBtucO8/sNIYQwgMHu\nFD4GzLc9BXgfcAZwMjAtHxsB7CxpbeAAYAtge+BYSWOA/YDb8rkXAofl1z0L2M32lsCmkjaStDEw\nxfamwK65rRBCCAUNlhQuBf6n5dxngY1tz8rHrgS2ATYB5th+1vZC4B5gQ2AycFU+9ypgG0njgDG2\n78/HZ+TXmAzMBLA9DxgtaY3lfH8hhBBehAGTgu1/2F6cP8gvJV3pt37PImBVYDzwRD/HFw5wbCiv\nEUIIoZABxxQAJE0EfgKcYfsHkk5o+fJ44HHSh/y4luPj+jje17HW11jSz2v0a7XVVmH06FH9fr2r\na+xA3z4kq68+lgkTxg1+YhtjqEscdYihLnHUIYbljaMOMdQljjrEUJc4Bhtofg2pS2d/27/Oh2+V\ntJXt64AdgGuAm4BjJK0IrARsQBqEnkMaOJ6bz51le5GkJZImAfcD2wFHAM8DJ0g6CZgIjLS9YKD4\nurqeHPDNLViweMCvD8WCBYuZP3/Rcn3/cKhDHHWIoS5x1CGG5Y2jDjHUJY46xFAyjoESxmB3CtNI\nXTj/I6kxtnAgcFoeSL4DuCzPPjoNmE3qXppm+xlJZwIXSJoNPAPsnl9jX+BiYBQwozHLKJ93fX6N\n/QeJLYQQwjAbMCnYPpCUBHrbuo9zpwPTex17Ctilj3NvBDbv4/iRwJEDRhxCCKFtYvFaCCGEpkgK\nIYQQmiIphBBCaIqkEEIIoSmSQgghhKZICiGEEJoiKYQQQmiKpBBCCKEpkkIIIYSmSAohhBCaIimE\nEEJoiqQQQgihKZJCCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQgghhKZICiGEEJoiKYQQQmiKpBBC\nCKFp9FBOkrQpcJzt90j6Z+B8YClwOzDVdrekvYF9gOeAo21fIWll4CJgArAI2NP2o5I2A07J5860\nfVRu53Bgx3z8INtzh/G9hhBCGMSgdwqSvgicC6yYD30DmGZ7CjAC2FnS2sABwBbA9sCxksYA+wG3\n5XMvBA7Lr3EWsJvtLYFNJW0kaWNgiu1NgV2BM4brTYYQQhiaoXQf3QP8BykBAGxse1Z+fCWwDbAJ\nMMf2s7YX5u/ZEJgMXJXPvQrYRtI4YIzt+/PxGfk1JgMzAWzPA0ZLWmN53lwIIYQXZ9DuI9s/kfSG\nlkMjWh4vAlYFxgNP9HN84QDHGscnAU8Dj/XxGq3HlrHaaqswevSofmPv6hrb79eGavXVxzJhwriX\n/P3DEUNd4qhDDHWJow4xLG8cdYihLnHUIYa6xDGkMYVelrY8Hg88TvqQb41gXB/H+zrW+hpL+nmN\nfnV1PTlgoAsWLB7w60OxYMFi5s9ftFzfPxzqEEcdYqhLHHWIYXnjqEMMdYmjDjGUjGOghPFSZh/d\nKmmr/HgHYBZwE/BuSStKWhXYgDQIPYc0cNw81/YiYImkSZJGANvl15gDbC9phKTXAyNtL3gJ8YUQ\nQniJXsydQnf++2Dg3DyQfAdwWZ59dBowm5Roptl+RtKZwAWSZgPPALvn19gXuBgYBcxozDLK512f\nX2P/5XtrIYQQXqwhJQXbD5BmFmH7bmDrPs6ZDkzvdewpYJc+zr0R2LyP40cCRw4lphBCCMMvFq+F\nEEJoiqQQQgihKZJCCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQgghhKZICiGEEJoiKYQQQmiKpBBC\nCKEpkkIIIYSmSAohhBCaIimEEEJoiqQQQgihKZJCCCGEpkgKIYQQmiIphBBCaIqkEEIIoSmSQggh\nhKYh7dFckqSRwLeBDYFngE/ZvrfaqEIIoTPU8U7hg8AY21sAXwZOrjieEELoGHVMCpOBqwBs3wi8\ns9pwQgihc9Su+wgYDyxsef68pJG2l76UF3vyiUdeciDL873D+Tp1iKMOMdQljjrEMFxx1CGGusRR\nhxjqEMeI7u7u5XqB4SbpZOAG25fm5/NsT6w4rBBC6Ah17D6aA+wIIGkz4PfVhhNCCJ2jjt1HlwPb\nSpqTn+9VZTAhhNBJatd9FEIIoTp17D4KIYRQkUgKIYQQmiIphBBCaIqkEEIIoamOs486Qq7xtAmw\nIjAC6LY9q9qoqiHp8F6Hum0fVUkwFZK0ju2Hq46jQdLqQJftSmajSHq/7V+2PN/F9o8qiqXqf4vx\nwA7ASvlQt+0L29FWxyUFSduS3vdI4FvAV21fXEEoPwYmAPNajhVJCpJOs/3Z/HgP2xflx5fb/lCJ\nGHr5O9BN+j/ZmAruYCW9DXgVsBT4OvB121cXDuM3kg6xfXnhdpchaSvgDGAU8CNJf7b9nYLtv59U\n7mZ3SVuQLppGAjsDRZNC1f8WLX4G/JVlPy/aouOSAnAMsBupEutk0g9ZFUnhNbnoXxX+teXxJ4GL\n8uNXVxALts9qfS7pqgrCOAuYChwFHAqcAJROClsB50raETjQ9pOF2284OsdyGakg5bVAyQ/C24A1\ngacAk5LC88APCsbQUPW/RcMI23uUaKgTk8KTwCPAs7YflvSSaioNA0t6re2/VtR+bUh6U8vTdYHX\nVxDG08AdwAq2r5f0XOkAbP8F2EHSx4HZkmbQ07U4rWAoS20/JgnbCyUtHPxbhtXfbJ8v6UekZNBQ\nRddN1f8WDb/PFR5uJf872F7SjoY6MSksJFVhPVvSVFKCqMKWwIOSHiX9J3fbXreiWKp2Dj2/8E8D\nB1cQQzdwIfArSbsAz1YQA5LWIpV5WUDPVXLpD8N7JB0HrCHpK8CDhdu/kHQ3fwfLvvduYFLhWKr+\nt2jYGvhAr2P/1I6GOm5Fs6SVgEm275D0VuBu289UHVdJkv5M6jIaAexO6j4bAexue70K41oNeN52\n8asxSRNIA/9Xkn4Bb7O9oHAMuwLHAyfaPr1k273iWIHUrbghcCdwdruuSl9ETKNsPz/4mcPe7mhg\nb1KXay3+LdqtE+8U1gSm5SuyS0iDizeWDkLShqS+yYnAw8Anbf+uUPNfzX+PAA6n52rsfwq1D4Ck\njYHzSB/GHyD163dJ+oLtn5eMhbTL3zjgv/LziaQr1pIOBLa1fVfhdnt7B7Ci7f0lfY9UpLLUz2aT\npD1I3UcrAidIOtH2iYXDuIx0J3v2Sy3fvzwknWF7qqTre32pu11jkp2YFM4hDRh9lZQMvgNsWkEc\np5G2Gr1N0kakge9SA8/bkH7If1Oovf6cBOxp+1lJx5Cm3N1N6t4rnRSKze4YwGRgI0nrN7aglbQy\ncKTtLxaM43Rg1/z4cOAC4N0F2284EHgf6eLt9cBMoHRSOJpUlPNYSZcD59n+c8H2G1Ozd6NQN2In\nJoWVbV8j6TDbt0t6qqI4Rti+DcD2/0kq2Yf9E+CLkr5NSooX2H68YPsNI3NSfC2wiu1bACoa/C82\nu2MApwEbAa+WdDQpQV5K+VlQS2zfA2D7PknFu22yxu/mQttPSxpVOgDbNwM3567Ns0j/JysWbP/v\n+eFrSIm6uU4B2L8dbXZiUnhK0vuAUZI2Jw1sVuF5SR8grU2YQuq+KCLPg79c0tqk7pL/J+mPwDm2\nZ5eKg57B3O3JH3y5P3tswRgais3uGMBk0jqN8aR/j3HA3ravKRzHnyV9HbiB1LVX1Qy5e0l38wfl\nBY7F91aRNAXYE3gXKUEfUjqG7ALgOKBx8da2u4ZOTAqfJt2CrkGa5bJfRXH8N6n75FjSANbepQOw\n/TdSX+2ppPGEqyl4FQRck/fNeD2wk6RJpIVCVaxa3ZplZ3dUMdPl8bxi9om8gnV72w8UjgFSd8m+\npO68O0ldKMXZ3kvSWNuLJd2cf15LOxA4l9TVW+WsnLtsn1+ioU5MCk8B37H9v3lKalfJxiWtYPtZ\n4CHgYy1fKv4DJ+ndwMdJ/cU/Bd5csn3bx0n6OfCE7b9KWp90t1J8Ra/tDaE5JfSxKma69PJw6YQg\naRPbc0l3rnfkP5AWb80sGUuO5+3APnnGIJK6bf93obYb/xbn5EPbSmqsGSn+bwH8WNIlwB/pWbvS\nllIwnZgUfgicmh8vIE3NfH/B9htzsO+iojnYko4kTUW9m3QVNDUnquJs39Hy+F5Sl0Fxkt5DGl9Z\nSOrT36eCX/71c7fNCOCfWh6XWrz2XmAuPYPMrar4IDyfVIrmL/l5yQunuv1bTCWVxnmcNq9d6cSk\nsIrtXwDY/oGkfUo2bnu3/PAj+UoEAElbFwyjG/i3wrMo6u5oYEvbD+WB78sp/8v/P/T8srdOFS7l\n30jrJB60fUThtvvysO3pVTRs+/j88Le2z20cl3RgFfGQ7l6PH/y05deJSeFZSdsB15MGj4p2E+Qu\nmzcDn5P0jXx4FPAZ4C2FwngYeJ8kSFcdDd22z+n7W17xnrP9EEDuyqpiVtrMRgyt8gB4CWtKugx4\ndy490vjZ6La9e6EYWj0g6cukwf9GHEUStaTdgJ2A90p6bz48krSI7dR+v7F9HpV0Nj3rRdr2u9qJ\nSeFTpAHeU0mDaJ8u3H4XsA5patk69BT7+kLBGNYu2NagJL2ONLOisaDwdtulFxQuknQAPbPBiq5m\nzi4G3gPmFwy9AAAgAElEQVQg6Xu2P56PH9s43mb7A2OA9YGzaUkKBdruy0qA8p+GUndvV5EuntYk\nTUVt/J5W0r2Z2+2mwO9uJyaF/WzvXFXjtm8Hbpd0TutVoaQxBcOYa/tXBdsbTB0WFO4BHEaqonsn\naXZYlV5XQZun2t5U0sO2r6ug/WXY/kTrc0nFaoPZ7iJVRL02tzualBjWI00SKULSeqRk8N1SbXZi\nUnizpNXyf3qVdpL0eXr2dljEsiWt2+kLQJ2SQmULCiVNtD2PtDjo3JYvTaDwzLQauE/SI8Cqklo3\n+6mkWKOkr5Gmxq4IrALcDJTqSmvEcF5ucyywMumipeTElEYNrIk5hrnA24FHSUU1h10nJoUNSP1z\nj5I2VKmqOulU0tz4Q0n1VXpXQGynkf3dmVRU7KvKBYWfBz5H6i7pPRvsvX1+xytUYxJEo95O1fGQ\n+vQnAt/If75cQQxvA95K6kI6lMLjCbY/ACDpF6TJKU/n392ftqvNjksKVVYB7eWhPNNlvO1f5wG1\nUjYllWXurYoFW5DGdU4i9d8eQsEFhbY/lx+e3JiVBiDpo6ViaDG55Qp9jZbHq5cKQNIbgPmSvksq\nE31+RQvoIM0+ejr/jtyTu1JKe8z20ryIbn6uAlCFdWw3LpaeI42/tUXHJYX8w97NsjMrqug/fkLS\nh4ClkvYlbS5TyvW2SwxcDtWHSWM9xQd31bP14275LqV168dLSsZiu+S40gtIehdpPOcM0uy8NwJX\nSPqk7RsqCOkvkj4JLFba02BCBTHcIukLwEOSfkg1JVgg/T/MAm4hXdRd2q6GOi4p0POLPoJUZ6aq\njW0+RZrlMY3UhXFARXHUwWjgfyX9CTjX9rUF225s/fg01W/9iKRtbF8t6cQc11LgK7ZLbAZ1NPDv\nLetXZkj6FWkiwDYF2u/ti6T6T5cCnyAtuCzK9lckjSNVQtgBuKl0DDmOwyX9jDQT63zSmEJbdNwm\nO71J+l/b21bQ7kzb25VuN7f91jwLqlYkbUIaBN/I9psGO3+Y2x7ZWi9f0jq2Hx7oe9oQw2HAv9r+\nqKQbSAvYpgDjbbf9okHSr/u6g5Q023bx0tmSfmO7LYOpQ2h7b9vnSjq215dKb426jLxmYiow2XZb\nurI67k5B0vb0DCiuSxv75gbRJWln0tXpUgCX21zlW3nhGqR/i6dIV0AnuoLN4pX2DPhPUsXWxsY/\npR2Zu/Eqm+kCbEtaVQzwlO0Zkq6m3NXpC0pT53o/VXVrLcgriE3PlrWl1ik07pb+lP+ubM2GpLGk\nSq37kdYpfJZl66YNq45LCiy7WcXTVDcf/TXAQS3PS8526T2QO5Z0a3wWPTuPlfR7Ul2X/Rp1/CtQ\nh5ku2H4uPzw1P39e0hOFmr86991Py4Oro0jrNqqo9QNpAeFG+U9DkVhsz8gP39U6E0tpJ7oLSsSQ\n2zud9LlwOfAh4DTb329nmx2XFGx/Qmlv5jeT9me+dbDvaZPKZrvY/lMfh2+W9NtSMcAyFWPfTtpb\nobsxVbaCqbF1mOmygqQxtpfY/ik0FzWW2lzmWNJOXw9IWkCa9fQj0lTMYiT9yPYuvRevFY7hM6T3\nvbqk/8iHR9BTObaULUl3rTcA95VosOOSgqTPkgasbgAOkXSpC+77WqfZLn0o/fPQqBj7h17Hq5ga\nW4eZLt8HzpP0WdsLlHb7OiUfb7ucoL8CfEWphPjjFa1bqeLffhm2TwdOlzTN9tcrjGMjSZNJE1O+\nQVpjtIHtO9vVZsclBVJC2NL2c0q7fF1P2X1fK5/t0sfCtZVI00JLzHBpaqkYu0uFFWMb6jDT5XSl\nrS+vk7Q6aZX7GbbPriCWoj8LvUxST9nwVsUGeSV9IN/JP6aeSsqNMuZFi0bangPMUdp46WPA9yRh\n+53taK8Tk0Kz39Zpw/iiV0K5pML5ki4k/ZCNALag7G1p770cniJVXyxaRlz1qBjb8POWmS6nFW67\nyfaZwJmtxySt7Wp2HavKk/S9uLKkxoLBdaiuIOAybC8k/WycKWnjdrXTiUlhjqQfA7NJ/XVzKorj\nm6TCa+uR+tT/Tpph0Ha231CinSHoXTEW0kyskhVjGxozXe6ip/xJ0QFWSe8AjiANsB6SV9DuTyrU\nV8l6mpbaUCX9zXaxwdy+NNq3fYSk15B+RmvD9u8GP+ul6bikYPvg3K//L8B3bV9RUSib2D5Q0rW2\nt5ZUbHP2fBt6FvBp24sk7U6afbO37UWl4mitGAu8xvatkj4I/G+pGFpUNtOlxbmkWU9vAL4u6VWk\nZDClZBCSvkja4evVwCckzWgpB1LCLQXbGpCkbwM7kspoQ7pr2KKCOJZJzpJkuy13Ux2XFJRq999P\n2oryS5L+avv/KghlZL4yvF/SiqT+7FLOJlV7/Ed+finpw+dMUgnp0r4F/JK0mcobSVP+ivbp51lp\nb8rt/x74a8n2s8WNuxNJXyUNxH/M5TeM/zBp3+4ZpG68/1eycduHlGxvEO8CJrUubCxJ0r+SfjeP\nz8kaUhfrcaRifcOu45ICaSbH4aR+68tIszu2riCOC0kfwnuRtkAsOZj4+pZB3sask5PyKtoqvNb2\nd3MsJ0q6tnQAShvsfJDUl/w90uynzxQOo3UXwIdsF50K2uI50iKpv9nuzosLO9W9pJLZ/xjsxDZ5\nNWmG3mvy35C6N89oV4OdmBSWksYTDnXao/lTVQRh+9vAt/PTgwY6tw2e7ed4FdMPIRUFlG1L+mfS\nFN3SdiV101xt+xuSbq4ghkZJ8xFAo0QyUHzdxrXAdcDHJH0TqKSLtSYD7K8HHpR0Dz2rqot1H9me\nDczOA8vzbc+TtEnrbL3h1olJYQXSlfksSe+h8BJ+ST+2/WFJf6NX/X6X29fhXkk72/5ZS1w7UXBH\nqV4+B1ySB/QeovwWqZA+iFu7CEru6dCwHj2zbka0PC66biPfoRwKIOnmitYqAPxY0nxgOvCrirpw\ndq2gzb58mtTlfRIpWe9h+8B2NNRxBfFyv/E2pBLBOwM32y6yUrAu8qKoH5BuSe8nbf34KPBx249V\nGVtVcvfRLqQP5j8C19g+qdqoqiFpO1Kibsy46bZdyYZDkt5CWjfybuAa4Dslfl8l9XdhUnydAoCk\nubY3aXnetiKFnXincB+pm+RQ0g/ZwiqCkPR20rqA1l+8InWYnLYifV8u5bAu8KBb9osuTdL9vQ49\nYXujPk9uE9vfysXn3pqe+vcl24fmXh8N3aT5+nOB7xW+Sv4mcCDwl4Jt9uevpN/Zd5L+b74p6U+2\nv9TmdmuzPiHrlrSm7UfzRV3bSp90YlI4m/SDth1pwdaFpClnpZ1PmnXT+MUr9gMo6b9tn2f7QUnj\nGglB0hG2jygVR4sN8t+NPS4+UjqAfAd5HPAm4A+SDq5gfv4lLLsB1FjgfaR1LCXHnR60fXXB9vok\n6UekfcsvIs3Cavyctn28p6Lfg4EcBcyV1EUafN6/XQ11YlJY3/YnJb3b9k+VdlWqwsO2p1fU9seB\n8/LjbwGNGvpbVRGMe7YZhLS48LgKwriQtMnMHNI89PPpKWNdhO2r+jh8maQbS8YBPCLpLNIUYaio\nywT4he1dGk9a5uYX39uharZ/KelKUomcR9o5TbkTk8IoSWsCKO2oVMn8Y1Ilyi+z7C9eVSWKK6Vl\nNzJZh2WnZpbyD9u/zI+vkPT5CmLoT+mf0QdIdyyV7EfcMjf/YEl/z4ebc/NtP1VFXFWS9Otez9s2\nztOJSeEw0tXgOqQFXG0ZwR+ClUhb66nlWEcmBdKsikYi+D+gryvmdrtH0lTgatKCpcWN+jLtLCkw\nmDxDruhMqFzaYV3STL0RlC+xsRqF5+b3J0+Rfpft7+eLl7NtP1A6Dnr2QGl0sb69XQ11XFKwfZ2k\nfyGV551P+sErRj17CFQx7bJh9TzDZETvxxXFs5sr2BK1lxWBQ4BNSFfJC+jZN3uvEgH0MU15KWl2\n2N4l2m+J4zzSrnNjSQu3bgTeX6p927NIU8Y3rjIhZxcCB+fHV5JmLRbtVoQX7IFyZzvXV3VMUuij\n2NgjFRUba+wh0NhisKHkXPRb6bkCa31c1S/gAlW0NanSVoc/IPXV3kAq6zCflKhK7XgGgNu05+5L\n8DbSTJ+zSLP0Ti3ZuKQznHY7O0NqvZEuu3Cspc3rISUrSVUsrOw9RXYd4FXtaqtjkgI1KTbWKC9R\ncaXSaX1NQZVUyUAz1W5Nejxwqe0LGwfyVdgJFL6bk7RSbvM04LWkEixPky5iSq7sfcxpO86xuVJr\n6WT1tfx342KlcfHUe3+FEp5Q2k/helK3YrGCkb20/h88RVpT0xadlBTqUmyMHMO+pA+A1nUKby7U\n/FWSPmf7mhzLSFI9qN1IUzJLq2xrUtLA5dTWA7anV1T+5FvAYlKZj28DN5H22TiTtD9vKbfkWXkP\nSfohqRuppMdy//2RTlukvp9U5v6wwnFAKmd/GOnf/06q29P9X1rrlbVTJyWFuhQbazgQ2IFUori0\nHYAfSmpMvbwI+DPwjpJBqB5bk/ZXB+q5Qu23erPtybkA3ZbAh502gjp4sG8cTra/kmfmPUX6Wbmp\nZPukO6Rn6blDuB7YnrQd5WdLBNBSqnp1emqUkZ8/WiKGXsZIehvLdrG2pfxIJyWFuhQba7gN+Ivz\nLnAl2f6rpPcCPwO+CnzBdtF+46zyrUlJ4xnLFBiTtAlQRbmPxfnvLYCb8oQEKLzBi6RNSTV/Gu3u\nQBsXS/XhHbY3azyx/ZjS3uolk9PnSaU+zuaFC0vf88LT207AT1uet20MspOSQmuxMaio2FiL/wfc\nJ+neRhyl6svk/Ru+CaxB+mU/JK8LKjoV1PXYmvQQ4Ge5XPd9pDGnbYEPFIyhYXEeUPxP4Pu5W+9j\npLu4ki4grQlo3MWW7mJ9wTqEXMK7WPlq502FbG9dqs1B7OVC+5h3TFKoeGC3L/uSyjkUneGS3URa\nCzDZ9nO55s8lkv7NdhUrvKvcmvT+fGX878A/kf5tDrVdRf38fUlbkV5Jz4rq/8zHS7rL9vmF22w1\nv5+7t+L/J5L+SpoIMZ90V/sU6edzaonFpqpgH/OOSQo1NI9UobWK1buH2G5ueWn7gfzDV1VV0Mq2\nJgXIK2QvK9lmP3HMB77Ycujq/Ke0H0u6hFQtdgTpLvaogu0fTLp7+zPp7m0iKWEXr4kFzAKOsG1J\n65MmZHyNNA5XYrFp8X3MIylUZyXgNkm307N5R6ktKF/QX257iaSi2y62qHJr0vBCU4Efk7qPRlC4\n+8h5IxnSYPs6pO1ib6hopuDEXG8J2/dKWs/23ZL6m6AwrNyyj3lLQcDX225bl2Ikhep8vcK2TyYP\nlkm62vY2+fiBLDuYVUqVW5OGF3rM9vFVBpDvoK+rMobs4Vyg8Xpg8/x8W8rvUriHpMdJFVI/IWlG\nY9xjuHVcUpC0J2kRW+v6gGIDzbn9vlRVu71tddmHytVuTVp7kkYXnqX2mKSz6VnhXlWV1Dr4L9K+\nJ+8DbidVRXg7PQvrSvkwqTrsDNJYQtvu6jsuKQBfIs0sqWoDkQmkBLAdab7zLFKdmdVIV8wdp+pE\n/TJwBWmefil357/rUnYDSDsGOm0QVdKzpAHuR0lJYWyj7EVhz5H+P/6WZ2Kt3K6GOjEp3Gv7nqoa\nd97iUdL2tj+WD5+dZwCV0rpmY5nHBWNoVXWirjXbJRMCFFw9OxBJp9v+TH68PXA68MbCYTQ25doW\nuIXqNuW6ltSd9jFJ3yRdKLRFJyaFpyRdRSrR3BjgnVZBHGs0rnwkrUXqKyylvzUbVak0UdeFpBWA\nnYAFtn+dj60NnGq7ZOmPYqtnB7FQ0vGkMhtvIXXhlFaLTblyBYZDIe08187/j05MCr+iHnuvfo1U\nY+YJYFV66qW3Xe81G5JGuuwewL3VJVFX7WJSd8U6ShvWPwBMJxXIK6n36llIU0KLsj1N0omkD+at\nS7ef1WJTrlze/nPkLtbYZGd4XUyqmV/VBiIA5KuOX5LGGB4lLVKqytWUq0ral7ok6qpNsv3O3J13\nC2mGy3ts31kyCNtvLdleb33sK/EaSQ+TLhZK/74eBvyW1J9f5aZc38xtt72LtROTwuWk9/06Uh/6\n74DvVxTLBNIdwn+T6gCVrPnTqoqSxK0uIlWMfQupy+KsasOpzEJorhkZCWxre0Gpxhv7GEjqPZBa\ndB+Dxr4Skt5ru6q1Mw1P2n6TpAmk9T1FS+23eNB2kXHHTkwKa9reTNJ0UsXFi0oHkPct+AxpattS\nYPNcB6gqv6mwbYBzSAulZgJbk/a++K8qA6pIa3J+pGRCyBqrlnejHnduR9DGqZcDqaK8xCAekXQW\ny3axtmWacCcmhX9IGkGaWvZko7+wFEm/IxV8Owe4BriidEKQ9Ppeh6Y3jrVzpeQA3mj73fnxT/u4\nUu0Ub5H0fVJyeLOkxp1jkdXutv+eH65AKikxmnQ3vQ4VbR8r6af0DHiXHGvqXV6iUcG3koFm0vhS\nNwW2D+7EpHA5qVz0bZJuoHyRrRtIewi8jzTVrQo/4oVXgv8MjCftVVzaipJeZfsfklahuqmxVWvU\n9hnBsqu6S1+1fx/4CanMxENUs38AwHlUdMfSV3kJgNaS+4XjOULSv5PuUu6y3bbKAx2XFGyfLmlE\nXgDyS6DoVEjb++cPvl1IdwtvkbQfcEmp7oLWWvX5h/xrpD1fq5jyB2kP4P+T9EfSLfvhFcVRtQ/b\nLrKJzCAW2z5W0pts75V/T6pQh7GmnSR9np67pkXAv5YOIpfaeCMwG/ivPEW2LZsvdVxSkLQTsJfS\nfriQrkSKLkax/SSpNPL5kjYAPgX8gbQvbzGSNspxXE2qVPpMyfYbbF8saTbp1vjvFXVh1UHxD5t+\nLJW0DjBWPXuZV6EOY01Tc9uHkirpVrHPBsCUxmC/pFNJM6HaouOSAqk89D5Usw3mC+TphgdL+kqp\nNiWNAr5C2sBlH9uzS7XdTzxHAGPyvPRLJN1q+7gqY6rIukqbxPeeDVa69tBRwAdJV+r3UcFkjKwO\nY00P2X5I0njbv5b05QpiABgtaVQuFDiSNq6X6MSkcLvta6sOorfCK0Z/S9ph7ATSgOab8/GqCp/t\nZHtjANsflfRb0s5fnWYMPTXzK2P7OnoqlP6swlDqMNb0hKQPke6e9qW6u6ZLgDl5HHRT2riHeScm\nhZ/lf9jGgqBu2/9dRSCSViWVnLjP9uLBzh9GV+a/x+Y/UEHd/BbPS1rR9jMtdZg60QO2j6w6CEmH\nk6ZeNiqzFl001hjzox5jTZ8C1gemkfZtPqCCGLB9sqSZpNXm0/NAeFt0YlI4kFSzv7ENZiUfhJL+\nk9RPORq4VNJS20eXaNv2ESXaeRHOAv6QNxz6F9L/TyeqajZabx8AXp93pKvCbEkfy2NNV5H2UL/f\ndhWzoC6zvV1+3JaB3YHkelhHAUfa/oOkN5D2VjisXeXUOzEpPGy7bbdeL8LnSZt2XEnacOcmoEhS\nyCUD+lJFGQFsf0fSz0m//PdW9MtfOdt7SFqz8f4lvR94utRK1haP0HOXUIXjgaskHW37YvrYKbCg\nLkk7s2xxwLsKtn8KqR5W4+L1t6Sy+98gLb4ddiO6u+uwcLEcST8mdZncSoXF1yTNtv1uSb+2/R5J\ns2xXtYS+UpLeStp5bTXgAuBO21VNg6yMpN1J04M3IE0EeB/wN+CWEneRLYvl3khatFXFVrGNWFYD\nTgRWJv1MNPaKLrEvcmsc17Jsb0LbCtH10/4NrVPI87ERwE22N2lHm514p/AL6rGE/zf5l/C1Srtc\nzS3VsKR+p/XZrmKjn9NI9Z/OIS2c+jnQcUmB1F/9NtKV4b7AO0hJ4XrK3EWeTfrdaIzpVPZ7kkvK\nzyWtIG6dhFE0KQAn2/5F44mkkiXMAV7QhZfXWLVt0W0nJoX7qw4AwPZXJO1AumO5s/UHr4DvAA+S\nEuTTBdvtl9Nm6Nj+q6SFVcdTkSdtL86zwR5xz0btz5do3Pa1kj4NnGf72Vz/5y22iy4akzSJtJr5\nUWCzKroTc9fdZGA3SZvTswnVzrRx5k8f5kvaxHbzolHSJrSxEkMnJoX96LkaatSsL75BuKSbST/4\nZ9ku/SH4WuCjwPuBeaRy4tfmGR9VWJCn+71K0m7UZA1JBboljSeVUb8SQGkDpiK/p3m9yL+S1iU8\nSyrT/HlJa9k+aqDvHWazgC/brmp9BKSqxWuSLppMT+2j0pWMDybNmPwzac3IRNLeFh8Z8LuWQ8eN\nKbTK0x8vtb1zBW2vDXycVO7ij6RpZsWrlUr6Z2B3Up2b39kuvjgnT82dRvpAuhM4poIKoZXLd45n\nkZLidqTpypcAny1xJynpJtKV+dKWYysA19t+Z7vbb2nzdbYr35o1992PsL00j3s9Y/vuwb6vDXGM\nIv1+rkO6w7+hnRdwnXin0GoF0oyX4mz/DThR0iWkAbVfkAZaS3ueNNNkVVJRvKJyldqnbH8pPx9B\n6k8/s3QsVbN9JSkRACDpGdKH9N/7/65htdi9duDL3UiLCrXfaLMOCWFb0p38pLzK/AvAo5LOtT29\nZCx5FXOx3oyOSwpadlenFUhTvqqI47+APUn/B98B9irY9jqkO5RdSH2TPyBt6FK0G0vSNNIA8wqS\nPgncDfyQtNlMxyUFSXvS87PZHOzNYy0lJgA8KWl92/e2xDSJiragrNjhwLtyUvwysC3wZ9KHc9Gk\nUFrHJQXnXZ1q4G3AZ1x4q8VsHnAXqWvib6TyCrsp7ftasszFrqSVqmuSksHawHG2zysYQ51swLIz\nfkYCnyDNQCmRFL4EXC7pGtKEjImkabF7Fmj7BfLCtZ8Al9ueX7j5JbYflrR+fnx3jqnIoH+VOiYp\nSPpuy9Nlpt2VLHMh6QO5f/guYEqe4dGYg13qA7l1emMjSVZR5uKxXPPpIUmvBT5i+3eFY6iN1vGc\n/GF0AWlq7kGF2v+jpCmkGTbrkLaqPcp20e6jFp8CdgK+m8f/rrB9aqG2uyWNBv4dmAEgaSxp3URx\nkrYAvk36fX2QVMjytna01TFJgXRV3EgGxwNfpJoPwtXz3613LEXjaJS5yB88awJ/sV11iYV5nZwQ\nWkmaCnwOOKj0Ij7bj5OSUeVs/yWvVViNVLX1o6R6SCVcSJr0MAZ4bx5o/h5weqH2ezsd2MP27TmW\nc4C27JvdkbOPGquIK47hq7a/1vL8uFIzf3L9lB+Rph3+nTS4+Q/go7b7K4HRjjgMnExKip9veVxV\ntdZKSXod8F1SWYf9O3EGVitJC0hXxccBM3LCKtn+qqQZR0/ncbh1qrpwkXS17W36ez6cOulOoRby\ngOqnSCWrG5v7jCRdkZSaDvpN4POtU2DzbIszgP8oFAOkAe51+njcqW4HniFtVn+GpMbx4mUmamJH\nYAfgk8B/Srqm5EI620+0PH4YKHbB1Ie/SDqRtCHWu0ilvP8jx/aT4WwokkJ5FwHXkCqkHk26Ml5K\numIvZc3eayJs/6+kLxWMoY7VWqv2wfx36+17lSXNK2X7Bkl/Ie0TvTtp0L2KLTnr4AHSz0GjDtIc\nenbqi6TwUvSairp6S6XQopVBnba8fEDSRbTMSSdtejOrUBj9VcAcVaj90IdcZmI86W5tPVLXyU8q\nWPFeC5JuJXWlXQ7sXnLcS9JnnPZz38z2DaXa7Y/tIyStS5pGPwJY1/Zv29FWxySFGk1Fbdg3/91a\nbqNUUlhd0nYsW/hsBD2D4KECkt4I/JS0kPE+4K3AlyXtbNuVBleNbYA1SFVbR6hn850SPivpAeAY\nSV9g2dmKpYvyIek80l1CYwbUjaQyNcOuY5JC3djerfG4UW6jYPO3Arv1OjaBtMFNcZI2Aza1faqk\n7wHf7NCZSCcDu9n+feOApO+T9hWvasP4Ku1O6lJbnTTzZxJpR7gSvkS6Y1uLF/6uFE8KpHVNbyV1\nnx1KG2dhVbHnaXihouU2bH/C9l629yKtHB4FvJPqVmqeDlyRHx9OuWmHdTO+NSEA5OTYqXdwu5Jq\nQD1u+xv09Ke3ne3L8/qlffLvyZeAT+XHVXgslyAZmxfyta3nI+4UKlJluY18Z7IbMJU022U88E+u\nbvvFJbbvAbB9XyesGu1Hf3tTd+pYT2MSRkMVZd4XS7qPVHrl1ZL2qaL7CLgld2M9JOmH9OytPuw6\nJilIur6fL3XbbssikIHYXlvSq2z/Q9K6jdr5hTxAmgK6h+27JF1ZYUIA+LOkrwM3AJtQn72KS/s/\nSVNtn9E4IGl/4JYKY6rSD0jjbOtJupI03lLa0cCWthur7i+ngu4jp/1XxpFKnuxA2r63LTomKbBs\nv2DlU/xy7foxpJLR35R0q+3jCjV/CrAH8AZJ36H6bsS9SAPvO5BWkRbZq7qGDgXOzXtL3EuagXQv\n0O9Oea9ktr+V6zC9JT1dtmutkOcaF2xOG0AVvXjKpcuPAo60vahl858r29Vmx61ozqtGjyMNIF0C\n3G77xgri+J3tjVue/7b0HYukrUkL6XYgjSd8z/btBdvfxPZcSdv3+lIlMzzqIpcTnwT8tQblR4rL\n1WJ764by28VK+iWp9tEsYArwXtsfKtj+GaTKA1+y/YykNYAjSL8jn21Hm1VfIVbhHFIpgTGkaV2n\nVRTH85JWhGYff3/9yW1j+1rbe5D2UfgLaWFdSY0N0Hft9af3bI+OIGmF3I222PZNwNslHZcLs3WS\nDUgz4Rp/3kzac6Tk7m8Ne5Du2I4BXk8q9V7SO2wflNc3Yfsx4LPA5u1qsNN+2ABWtn2NpMNycamq\n+tLPAv4g6XbSD/7xFcWB7S7gW/lPyXYb7/m3ts9tHJd0YMk4auQU0lVh4/b9emB74BukD4KOUHW1\n2F6xPA4cUrrdFi/4fLLdLSn2aB5GT0l6HzBKaUPuSjaut/0dST8ndRPc6wo2J6+a0n7MO5GqUDbu\nGkaSlu934rTUd9huTru0/Zikz9LGQcU6q7JabI3Mb3SzNg5I2oRUwLItOjEpfJq0GGhN0hXAflUE\nkTi/g8gAABAxSURBVMvfnkkqC3yBpDs78Af/KlKRsTVJd06NzdHvHeibXsGKXxXWUa9qse/q8Gqx\nBwM/k/Rn0ir3icA/AR9pV4MdN9BcF5L+HylBnUPqt/y57XdUG1U1lPZlHk+ak/4h4Be5S6ujSPoR\ncGIfV4VH2P736iIrS9Lj9FSLbVW8WmyuRfVFYF1S+ZE/NNbUFIxhFLAlqYrwg8AN7Sz30XF3Ckr7\nAn+RnquyogXxWtm+W2n/3b9K6siiZ9kPSX3GW5DuFj6U/3Sa4leFNdVaLbZ1AkYVV7DnkaZ/bk26\nczmPNAupGNvPk/aGLqLjkgJpdsu6tp+sOI4FeT76q3LfetENRGpmXdvfk/RJ21tLurrqgKpge16+\nM2hcFV5Km68K68j2tVXH0GKNPP63h+1Z+a72Fa0Tp6TeR0WDy718knQV+Cip7tAnqw2nUivkDUP+\nKGkCMK7qgKogaUfbz9u+Drja9vV5TGHfQb85tEu3pH+B5lhHf2Xn20LSZ/Lfxeo+deKdwoqkqaB/\nIN2OVrWr1Ra2m5va5P/8qvZ/rdoJpDu4zwMHAF8b+PRXrC8Av8qPLwUaW8Z+lM7dXKZqB5IGvTcA\nLgP2L9x+8RLenZgUjqcGZS6AQyRtSZqbPh3o2BkWeTvBxu5R/5M3EwmhDu4m7Zd9q6QPAn8o3H7x\nEt6dmBR+RyrrsFJ+XlWC2Ja0KOcvwOdsf7uiOCon6Wuk2kcrAqsAN1OwTHIIA7iYNAniVtJmPxeQ\n9nkowvblwOWSPmD7F5LWIpXRblsl4U4cU/gZacOSxhL6DSqK42hSeYk9gU9L6sSN2Rt2Is20uYj0\nf1Ks/lLNrCLpTZLU+3HVgXWw19r+LoDtE0lTU6vQKOE9E7g375zYFp14pzAi1/up2mjg3bafkzSD\n1IX0/YpjqsrDtp+WNN72PZLWG/xbXpGeAs7u43HVM+U62VJJsm1J/0x1F9LFSnh3YlL4fR7Jv5We\nyotLSjXe2DvB/7+9O4+xqyzjOP5tgZJadhdAgkBEfmCLokUFRaAIom1ZFCFU2ghSRAItq1AWFxqh\nEMQALmxVFg2CFJVFipGlBRdANJQQ4IcJiFYKtLSsFrqNf7xnZu6dzhRIes974H0+yeTee+7M3Kcz\nnfPc8y7PY5/cfcz2Iknn1RVDA82VdDjp3dA5pNagxbG9e3/HJW1acyih1/HAdZI2Bp4mbTjNobYS\n3iUmhd1p73fbRY2tMEljlKMAJP2y5aplGr2rTUrzDdLw0a+BQ6lxzLZJJI0klUVeCJxke37VZOd0\nYLOcsZWqKqu/Q+44gJclTaK3hHfHFqYUlxRsfwSgjgmbN6HoP3RJ/b3rWkLavPVIzeE0weXAFGBL\n4GxJw0hj2LvlDKpEkm6wfYDa2+ZCvgoI44EzSCW8H6WDJbyLSwqSRgE/I3/P1ZB27TZheXBTvNL9\nf1HSt4GrgUNK29HcELOq2/1t35szEKi3hHdxSYGG9FwNQNqkNpp0MrwrdzAN0HrV+rTt07NFEiZJ\nehL4vqSTW594p7+JLDEpZO25CnxG0rzq/kat92uOowkuJlVHXUfSx22fnzugzAa3dOF7rboP1LsY\nIgC9m8Y2pqZNY01RYlKobcKmP7aHvPFnFWM725+tmpPfBpSeFLYA3PK4+37diyEC7Gv7MEmTbedq\n2dujzhLeJSaF2iZswhtaCmB7aQnVJ9+I7S1zxxB67CzpB8CBkjahvebQaRniqa2Ed4k7ml8kjWWP\no/46JqHdoAHuh5DbaOAh0iZCA49Vt17VF3XQu23/DFhq+246+PdS4pVCNHRpjoHmV7I1PgoBwPYT\nwBNVhdJBwFO2/5UxpNpKeJeYFLI2dJH0HtLw1Z7A+qTmOncDZ9p+rs5Ycov5lTdH0gjbpdaDykLS\nOsCvSP3DnwS2ljQfGGc7R5fE2kp4l5gUcjd0uYq0/vy7wMvV63+RVPdoz5pjCQ0laQRwHmneaySx\nga1u5wLX2766+4CkiaTfSY5SF7WV8C4xKeRu6LKu7etaHr8IXCvp6JrjCA0jaQzphHMBqSfvEOAo\nUpOdUK+P2m77m7Q9vUoMOdRWwrvEpLCL7YOq+9/J8PrzJX2HtATzRdI6/dHAvFV+VSjBNNIbldOB\nE0m9NrYlXUXelDGuEi0d4Hit7ThbtJXwljSrUy9UYlL4sKQNbS/K9PrjSe/+TiElhJeAP5P6KhQl\n5ldW8jowjDSO/T/bYwAk5ToRlWyhpE/Y/lv3AUmfIC0HzaG2Et4lJoXtgAWSFgArqHmli+3FwA+r\nj9LF/Eq744FjSEOa+0iaSlr5kutEVLKTgBurd+RPkIoU7kV7heU61VbCe1BXV9TaqlNr6YK+Sitl\nIOlu2yttwJF0j+3P5oipKSQNJQ0hDQHOt/1i5pCKU/0OxgBbAf8FbrT9at6oOq+4pFCt6rgY2BC4\nEnjM9i01vv7jpCbcfYevumwXVcpA0g3AHFaeXxneMu8TQrFylPAucfjoIlJpi8tI65BvIs3q1+Uz\npIJan7Nda92lBor5lRBWbVZ1W1sJ7xKvFO60vYeku2yP6r6tOYa9geW2a904F0J4e6lGFk4glfyv\npYR3iVcKCyV9ExgmaRxpxUutbP+h7tdsophfaVdVi50KTLW9WNJYUhe6M2zHCqQy1V7Cu8SkcDhw\nGrAA2LF6XJuqGuh+rLwMc0aBHbYeZoD5FcosFX0BaX38iurxX4G9SSvVJucKKmRVewnvYpKCpDVI\nk5gv2z4lYyg/IS0znAm8Qu8yzL2BXLslc4n5lXYjbe/U/cD285ImA/dnjCnkVXsJ72KSAu1dvkZm\n7PI1op9lmDdK+kuWaDKyPV/SFODjQMyvpDLNbWx3SXrHL4MMAxpNGkIcS2/Z7kF0sLd5SUmhKV2+\nBkvataqJDoCk3YDixtAh5lf6mD/ALtpICoXKUcK7pKTQlC5fhwI/lHQN6Ze8glTk6oiMMWVR/QyO\nt/1s7lga4kTSVeO/SbtoNydtnDowa1QhmxwlvEtKCo3o8lX1Vd23+7GkdwErbL+WK6aMPg3cJuki\n4MoCJ9rb2P5PdWWwC7ApcD1wb+k/l8LVXsK7mH0KkpbQW0NmI6B7YrPW2keShpP6Qy8ilcOdDiwH\njrN9c11xNIGku0hd76YCe5B+HjOBJzI1Msmq2r/SRfubli7o3Jr00GyS/mR7l36O39u6KGF1KuZK\noUFdvi4hVQbdErgB2IY0wXgbUFRSALD9AjBZ0vuAr5DKmW8DjMgaWB7jGHgCMZJCmWov4V1MUmiQ\nQbZnA7MljeoeT5c00C//neyZ7jtVqeyfVh+lusj2P3IHERql9hLekRTq97ik6cCRtg8FkHQqLSfI\nUthu26FZ7XBeoyovXqLzgVpLroTGq72Ed8caNYQBHQHcbHt5y7G5pFVJRVEyQ9I1knYm7XB+RNLB\nuWMLoQlsPwl8ilT1YC3SRsZPVktVO6KkieZ5wIQoQtccku4mTTKvD1wK7ECagL+jU5NoTdZnMUSr\nWhdDhLKVNHz0LHCspAmkdo8dy7ThTRts+/Zq38jZtudCsfMrAH+uu2JvCH2VlBQW2d5H0peBayW9\nANxKWv4YTdHzeFzSdaQrhXmSziL1VFiQN6wQylXcnILt39j+JHAcaTfx5zOHVLKJwBXANFLV2FdJ\n9akOyxlURmfnDiCEkuYUptg+J3ccIQxE0kakfRonAtuR2sW+Bhxu26v40hBWm2KuFPomBElxhRCa\n5lJSfZsu4EfVx2RSC9kQalHMnIKkI2kvIXCCpPMBbF+WLbAQem1i+0JJ6wHbA1dXpbOH5Q4slKOY\npADsD2xAKicxCBhCKjoWQlP8r7rdFbinpRBeJIVQm5KSwhhS8+s1SeO2u9k+M29IIbSZJ2kaafHD\n9yWtS1oQMSdvWKEkJc0prKja180BZgBDM4cUQl9HAf8BzrL9W1JRwPcAx2SNKhSlmNVHrSSNIO1u\nztmrOYQ3JGl74Gjb38wdSyhDkUkhhCaTtAZwAOkKYWNguu3z8kYVSlHMnEI/q4+6dcXqo9AEkjYF\nvgFMAP4KrG1beaMKpSlmTgHYFjgZ2KTPR6xACk3xT9KquJG2J9DbHTCE2hRzpWD7eEnbAjNt3587\nnhD68XVSafXbJV0BrJ05nlCgouYUJL0XGGb7X7ljCWEgkrYCDgfGA/cBv7B9S96oQimKSgqtJA0F\nVth+PXcsIfSnmnAeDUy0vV/ueEIZikkKkoYDZ5GauFwDXE6qknqs7ZtzxhbCqkj6pe3xueMIZShm\nTgG4BDiD1ON0BrANsJhU9iKSQmiyWIEUalNSUhhkezYwW9Io289C0V2+QghhJSUlhcclTQeOtH0o\ngKRTgWeyRhVCRdLefQ5176tZL0M4oVAlJYUjgLG2l7ccmwtckCmeEPoaR0oEff2l7kBCuYqZaA7h\n7UDS+sAy26/mjiWUqaQdzSE0mqRjgAeBByV9IXc8oUyRFEJojkNIK412JvVRCKF2kRRCaI7FtpfY\nXgCslTuYUKZICiE0R2sF3/jbDFnERHMIDSHpOeB2UnLYA7izeqrL9lezBRaKUtKS1BCa7iB69yZc\n2nI83rmF2sSVQgghhB4xbhlCCKFHJIUQQgg9IimEEELoEUkhFEvSjyV9rYPf/wpJm3fw+29VFXkM\nYbWJpBBK1ulVFrvT2b+xLYAPdvD7hwLF6qNQFEk/APYBngWWAL8gJYdjSSfwvwNH235d0lzgDmAH\n4GXgENtPSToQOAEYWn1MtH2PpFnA88Bw4ApgKvBPYFfgH8C1wFhgGXAacBKwNXCi7eslbUxqBrU5\nqSvgqbbvkPQ9YLPqc7cApts+W9JDwFbAlbYndehHFgoTVwqhGJIOAHYEPgzsRzrJDgMmAjvb/hgw\nn3SyBng/MNP2R0kn9IskDQKOBMbY3gE4F/hW9fldwBzb29o+F3gaGG17YfXcf22PICWIKcCewHjg\n1OrrLwR+bnvHKr5LJa1TPbc9sBfwKWCKpPWAScADkRDC6hSb10JJdgdmVD01Fkn6HWmj2IeA+yQB\nDCFdLQC8ZPva6v7VwDTbXZK+BOyr9AW7kd75d7tvFa8/s7p9Cphre4WkfwMbVsf3BCRpavV4TdLw\nUBdwp+1lwHxJC4H1aS+LEcJqEUkhlKSL9qvjZcAawK9tHwtQvTNfs+X5boOBZZKGAQ8AVwGzgDnA\nMS2ft3gVr7+k5f7yfp4fDIyy/UIVy2bAPGB/4PU+/45ICKEjYvgolOSPwMGShlTDL2Or41+S9N5q\naOhiYHJ1fKOWFpmHAbcC25BO6NNISWE0KbF0az1ZL+OtVTu9EzgaQNJwUsJ5FwMngGXEG7uwmkVS\nCMWwfTMpMTxMGsp5DHgBOJN0Qn64+tRzqtulwARJc0jj+ceRTtQPAo8Cs4GHgA+0vEzryo1bgN9L\n2rKfcLr6uT8J2Kl6vV+RJrZfqZ7vb0XII8AGkq5a5T88hLcgVh+FMABJi20PzR1HCHWKK4UQBhbv\nmEJx4kohhBBCj7hSCCGE0COSQgghhB6RFEIIIfSIpBBCCKFHJIUQQgg9IimEEELo8X+rgW/wUEgx\nSAAAAABJRU5ErkJggg==\n",
452 |       "text/plain": [
453 |        "<matplotlib.figure.Figure at 0x10d582990>"
454 |       ]
455 |      },
456 |      "metadata": {},
457 |      "output_type": "display_data"
458 |     }
459 |    ],
460 |    "source": [
461 |     "df_pandas.groupby('department')[\"total_compensation_float\"].max().order(ascending=False)[:10].plot(kind=\"bar\")"
462 |    ]
463 |   },
464 |   {
465 |    "cell_type": "markdown",
466 |    "metadata": {},
467 |    "source": [
468 |     "There are definitely a few outliers here. Let's just quickly see the distribution for the highest department"
469 |    ]
470 |   },
471 |   {
472 |    "cell_type": "code",
473 |    "execution_count": 25,
474 |    "metadata": {
475 |     "collapsed": false
476 |    },
477 |    "outputs": [
478 |     {
479 |      "data": {
480 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAECCAYAAAD5OrxGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAEa1JREFUeJzt3X+M5HV9x/HnDvf7blgDDDamFGOIb0sNpWp7eiocEfEn\n2pImtWoV2qpV22A0xYKWtkYrKZFUoqUGVLBaTSRXq7EqBgnitVW0tkrVN6KlGmtxF4Gdnd275W62\nf8x36/YKO7Ob7/duxs/zkVyY+e7n+/2+MsO85vtrvzu1vLyMJKkcrWMdQJJ0dFn8klQYi1+SCmPx\nS1JhLH5JKozFL0mF2TRsQETsBq7IzHMi4mTgWuARwBTwssy8u9mIkqQ6rbnFHxGXMCj6rdWkvwD+\nJjPPBi4HHt9sPElS3YYd6rkLuIDB1j3AHuCUiPgs8BLgcw1mkyQ1YM3iz8x9wKFVkx4N/Dgznwl8\nD3hjc9EkSU1Y78nde4GPV48/ATyp3jiSpKYNPbl7hC8AzwM+CJwN3DFshk/u/87yca3jhi54x9Qc\nZz35jHXGkaSfWlPDh2zMqMW/cie3NwDXRcSrgfuBFw+bcYopuvMHhq7gwf4BZma6I8apX6fTPqbr\nH9Uk5JyEjGDOupmzXp1Ou7FlDy3+6nLNPdXj7wHnNZZGktQ4f4FLkgpj8UtSYSx+SSqMxS9JhbH4\nJakwFr8kFcbil6TCWPySVBiLX5IKY/FLUmEsfkkqjMUvSYWx+CWpMBa/JBXG4pekwlj8klQYi1+S\nCmPxS1JhLH5JKozFL0mFGVr8EbE7Im45YtqLI+Ifm4slSWrKprV+GBGXAC8F5ldN+yXgtxvOJUlq\nyLAt/ruAC4ApgIg4EXgb8LqVaZKkybJm8WfmPuAQQES0gPcCr2fVHoAkabKs5+TuE4HTgGuADwOn\nR8RVjaSSJDVmzWP8q2Xm7cDjASLiVOAjmfn6UeZt79o2dMyOqSU6nfaocRpxrNc/qknIOQkZwZx1\nM+dkGLX4l494PvUQ0x5Wd/7A0DEP9g8wM9MddZG163Tax3T9o5qEnJOQEcxZN3PWq8kvp6HFn5l3\nA3uGTZMkTQZ/gUuSCmPxS1JhLH5JKozFL0mFsfglqTAWvyQVxuKXpMJY/JJUGItfkgpj8UtSYSx+\nSSqMxS9JhbH4JakwFr8kFcbil6TCWPySVBiLX5IKY/FLUmEsfkkqjMUvSYUZ+sfWI2I3cEVmnhMR\nZwJXA4eBg8DLMvNHDWeUJNVozS3+iLgEuBbYWk36S+D3M/McYB/wxmbjSZLqNuxQz13ABcBU9fxF\nmfm16vFmYLGpYJKkZqx5qCcz90XEo1c9/2+AiNgDvBZ4eh0h+v0+vd48c3MPjDR+1642rZanJyRp\nI4Ye4z9SRPwGcBnw3My8d5R52ru2rfnz3vwc//6D+9nc/vHQZS0s9HjB3tOZnm6Psup16XTqX2YT\nJiHnJGQEc9bNnJNhXcUfES8FXgnszcz7Rp2vO39gzZ8v9A6yzBb6bBm6rP7yQWZnuywt1bvF3+m0\nmZnp1rrMJkxCzknICOasmznr1eSX06jtuRwRLeCdwC5gX0TcEhF/2lgySVIjhm7xZ+bdwJ7q6YmN\nppEkNc4zpJJUGItfkgpj8UtSYSx+SSqMxS9JhbH4JakwFr8kFcbil6TCWPySVBiLX5IKY/FLUmEs\nfkkqjMUvSYWx+CWpMBa/JBXG4pekwlj8klQYi1+SCmPxS1JhLH5JKszQP7YeEbuBKzLznIg4Dbge\n6AN3AK/NzOVmI0qS6rTmFn9EXAJcC2ytJl0FXJaZZwFTwAubjSdJqtuwQz13ARcwKHmAJ2Tm56vH\nnwLObSqYJKkZaxZ/Zu4DDq2aNLXq8Tww3UQoSVJzhh7jP0J/1eM2cP8oM7V3bVvz5y2W2L5909Bx\nK2NPOqnN9HR7lFWvS6dT/zKbMAk5JyEjmLNu5pwM6y3+r0bE2Zl5K/Ac4OZRZurOH1jz5wu9gywu\nHho6bmXs7GyXpaV6L0jqdNrMzHRrXWYTJiHnJGQEc9bNnPVq8stp1OJfuXLnDcC1EbEF+AZwYyOp\nJEmNGVr8mXk3sKd6/G1gb7ORJElN8he4JKkwFr8kFcbil6TCWPySVBiLX5IKY/FLUmEsfkkqjMUv\nSYWx+CWpMBa/JBXG4pekwlj8klQYi1+SCmPxS1JhLH5JKozFL0mFsfglqTAWvyQVxuKXpMJY/JJU\nmKF/bP1IEdECrgMeC/SBV2Rm1h1MktSMjWzxnwfszMynAW8B3lZvJElSkzZS/IvAdERMAdPAUr2R\nJElNWvehHmA/sA34FnAicH6tiSRJjdpI8V8C7M/MN0XEzwKfi4jHZ+bDbvm3d21bc4Etlti+fdPQ\ncStjTzqpzfR0e725h+p06l9mEyYh5yRkBHPWzZyTYSPFvxOYqx7fB2wGjltrhu78gTUXuNA7yOLi\noaHjVsbOznZZWqr3gqROp83MTLfWZTZhEnJOQkYwZ93MWa8mv5w2UvxXAu+PiNsYlP6lmblYbyxJ\nUlPWXfyZeT/waw1kkSQdBf4ClyQVxuKXpMJY/JJUGItfkgpj8UtSYSx+SSqMxS9JhbH4JakwFr8k\nFcbil6TCWPySVBiLX5IKY/FLUmEsfkkqjMUvSYWx+CWpMBa/JBXG4pekwlj8klQYi1+SCrPuP7YO\nEBGXAucDm4F3ZeYNtaaSJDVm3Vv8EbEXeEpm7gH2Ao+pOZMkqUEb2eI/D/h6RHwMOB74w3ojSZKa\ntJHi7wCnAM9nsLX/ceBxdYZaS7/fp9udG3n8rl1tWi1PZUjSio0U/yzwzcw8BNwZEQci4qTMnH24\nGdq7tq25wBZLbN++aeg4gMXevdye93DCCYeGjl1Y6PGCvaczPd0eOhag0xlt3LE2CTknISOYs27m\nnAwbKf4vABcDV0XEo4CdwL1rzdCdP7DmAhd6B1lcPDR0HECvt0SrdRx9tgwd218+yOxsl6Wl4Vv8\nnU6bmZnu0HHH2iTknISMYM66mbNeTX45rfsYSGZ+EvhqRHyJwWGe12Tmcu3JJEmN2NDlnJn5xrqD\nSJKODs96SlJhLH5JKozFL0mFsfglqTAWvyQVxuKXpMJY/JJUGItfkgpj8UtSYSx+SSqMxS9JhbH4\nJakwFr8kFcbil6TCWPySVBiLX5IKY/FLUmEsfkkqjMUvSYWx+CWpMBv6Y+sAEXEy8BXgGZl5Z32R\nJElN2tAWf0RsBt4D9OqNI0lq2kYP9VwJXAP8sMYskqSjYN3FHxEXAjOZeVM1aarWRJKkRk0tLy+v\na4aIuBVYrv6dCSTwwsy856HG/8P+7y63Wmt/v/Tm5/jOf/4XZ/zC44au/0f3/IBWazMndU4eOrY3\nP8e5v3Iq09PTQ8dK0phpbKN63Sd3M/PslccRcQvwqocr/RXd+QNrLnOhd5DFxUNDxwH0eku0WofZ\nun342IXeQWZnuywtDd+x6XTazMx0h4471iYh5yRkBHPWzZz16nTajS3byzklqTAbvpwTIDPPqSuI\nJOnocItfkgpj8UtSYSx+SSqMxS9JhbH4JakwFr8kFcbil6TCWPySVBiLX5IKY/FLUmEsfkkqjMUv\nSYWx+CWpMBa/JBXG4pekwlj8klQYi1+SCmPxS1JhLH5JKozFL0mFWfcfW4+IzcD7gFOBrcBbM/MT\ndQeTJDVjI1v8LwFmMvMs4NnAu+qNJElq0rq3+IGPAjdWj1vAofriSJKatu7iz8weQES0GXwJvKnu\nUHXp9/t0u3Mjjdu06RBzc72RlrtrV5tWq77TI/1+n/n57khjt2zp0+9T6/ollWUjW/xExCnAPuDd\nmfmRYePbu7at+fMWS2zfvmnoOIDF3hZarc0jjr2X2/MeTjhh7Z2S2Zl7aB23iRNOOHHoMhcWerxg\n7+lMT7eHjh3VAw88wE3//H127Ng5fP13fL/29Teh0xnvfCvMWS9zToaNnNx9JHAT8JrMvGWUebrz\nB9b8+ULvIIuLh4aOA+j1lmi1DrN1+6hjj6PPljXH9Zc3sXP7zqHjBmMPMjvbZWmpvi3uubku/eVN\nI61/xw5qX3/dOp02MzOj7cEcS+aslznr1eSX00a2+C8DpoHLI+LyatpzMnN4E0uSjrmNHOO/GLi4\ngSySpKNgfI8XSJIaYfFLUmEsfkkqjMUvSYWx+CWpMBa/JBXG4pekwlj8klQYi1+SCrOhm7RJ0k+D\n9dwZt+678h5LFr+kYs3Pd/nsF+9i+5A74y4u9Hjm7tM4/vjpo5SsWRa/pKJt37GTHTvLuk3zT8d+\niyRpZBa/JBXG4pekwlj8klQYi1+SCmPxS1JhLH5JKsy6r+OPiBbwV8AZwEHgdzPzO3UHkyQ1YyNb\n/L8KbMnMPcAfAe+oN5IkqUkbKf6nAp8GyMwvAk+qNZEkqVEbKf7jgblVzw9Xh38kSRNgI/fqmQNW\n39iilZn9hxu8ODfDQu/gmgtc6M1zcLHLQm/4XfIOLPZotTbVOvbAYo9Nm+Bwf2roMhcXenS7c0PH\nrUe3O8fiQm+ksa2pQ3S7432LpS1b+szNjXbHw2PJnPWaxJyjfvZG/XxOiqnl5eV1zRARFwDnZ+ZF\nEfFk4I8z83mNpJMk1W4jm45/BzwzIvZXzy+qMY8kqWHr3uKXJE02T8pKUmEsfkkqjMUvSYWx+CWp\nMI1cEH6s7ucTEbuBKzLznIg4Dbge6AN3AK/NzOWIeAXwSuAQ8NbM/GREbAc+CHSALvDyzJytLlf9\ny2rsTZn5lmo9fwI8t5r+usy8fcR8m4H3AacCW4G3At8cw5zHAdcCjwWWgd9j8D6OVc5VeU8GvgI8\no8o3djkj4l+AB6qn3wXePqY5LwXOBzYD7wL2j1POiHg5cGH1dDvwi8DTgHeOS8ZqvhZwHYPPUB94\nBXB4XF7Lprb4j/r9fCLiEgZltbWadBVwWWaeBUwBL4yInwH+ANgDPAt4e0RsAV4N/Fs19gPAm6tl\n/DXwm5n5NGB3RJwZEU8AzsrM3cCLgHevI+ZLgJlqPc+u5n3HGOZ8PtCvlvdm4M/HNOfKl+l7gF6V\na+ze94jYBpCZ51T/fmdMc+4FnlJ9bvcCj2HM3vfMvGHldQS+XOW4fJwyVs4DdlbLewtj9hlqqviP\nxf187gIuYPCCAjwhMz9fPf4UcC7wy8D+zHwwM+eqec5Ynbf677kR0Wbw5fUf1fTPVMt4KnATQGZ+\nH9gUESeOmPGjDP4nhcFr/+A45szMvwdeVT19NHAf8MRxy1m5ErgG+GH1fOxeTwZbpTsi4jMRcXO1\n1TaOOc8Dvh4RHwM+AXycMX3fI+JJwOmZed2YZlwEpiNiCpgGlsYpZ1PFf9Tv55OZ+xjs5qxYff+F\nLoMX/3h+srt95PS5NaaNsoxRMvYyc756Az/K4Ft89esyFjmrrIcj4noGu9AfYgxfz4i4kMEe1E3V\npKlxzMlgb+TKzHwWg8NmHzri5+OSswM8Efj1KuffMp6vJ8BlwJ9Vj8cx435gG/AtBnukV49TzqbK\neF3382nI6vUdD9zP/8/VfojpDzVtlGWMJCJOAT4HfCAzPzyuOQEy80IgGByr3DaGOS9i8FvktwBn\nAjcwKK9xy3knVdln5reBe4FHjmHOWQbHjQ9l5p3AAf5vgYxFzoh4BPDYzLy1mjSOn6FLGGzJB4P/\nNz/A4LzJWORsqvj3MzjRQLVb+7WG1rOWr0bE2dXj5wCfB74EPD0itkbENPDzDE6y/G/elbGZ2QWW\nIuIx1e7aedUy9gPPioipiPg5Bl9qPx4lUEQ8ksEu2SWZef0Y5/yt6iQfDHZZDwNfHrecmXl2Zu6t\njvf+K/Ay4NPjlpPBF9Q7ACLiUQw+mDeNYc4vMDj3tJJzB3DzGOY8C7h51fOx+wwBO/nJ1vl9DC6k\nGZucTd3m8Vjez2flHhRvAK6tTpR8A7gxB2fQrwZuY/Cld1lmHoyIa4AbIuI2BlevvLhaxspu+XHA\nZ7I6U16N+6dqGa9ZR7bLGGxBXR4RK8f6LwauHrOcNwLXR8StDLZSLmawyzpur+eRlhnP9/29wPsj\nYuX47kUMtvrHKmcOriY5KyK+tGreu8ctJ4MrZVZfJTiO7/mVDN7z2xh8hi5lcOXZWOT0Xj2SVBh/\ngUuSCmPxS1JhLH5JKozFL0mFsfglqTAWvyQVxuKXpMJY/JJUmP8BtabKqR8ouaQAAAAASUVORK5C\nYII=\n",
481 |       "text/plain": [
482 |        "<matplotlib.figure.Figure at 0x115021cd0>"
483 |       ]
484 |      },
485 |      "metadata": {},
486 |      "output_type": "display_data"
487 |     }
488 |    ],
489 |    "source": [
490 |     "salaries = sqlContext.sql(\"\"\"\n",
491 |     "        select total_compensation_float \n",
492 |     "        from df_cast\n",
493 |     "        where department = '3147 Personnel Human Resources' \"\"\").collect()\n",
494 |     "sns.distplot(salaries, kde=False)\n",
495 |     "sns.despine()"
496 |    ]
497 |   },
498 |   {
499 |    "cell_type": "markdown",
500 |    "metadata": {},
501 |    "source": [
502 |     "Wow - someone is making a killing in this department. Or maybe an error?\n",
503 |     "\n",
504 |     "Spark also has some nifty functions. For example, we can quickly make cross tabulations. Check out the API for a lot more functions available."
505 |    ]
506 |   },
507 |   {
508 |    "cell_type": "code",
509 |    "execution_count": 26,
510 |    "metadata": {
511 |     "collapsed": false
512 |    },
513 |    "outputs": [],
514 |    "source": [
515 |     "from pyspark.sql import DataFrameStatFunctions as sf"
516 |    ]
517 |   },
518 |   {
519 |    "cell_type": "code",
520 |    "execution_count": 27,
521 |    "metadata": {
522 |     "collapsed": false
523 |    },
524 |    "outputs": [
525 |     {
526 |      "data": {
527 |       "text/plain": [
528 |        "[Row(paid_by_another_entity_fiscal_year_end_date=u'Y', 2011-03-31T00:00:00=7, 2011-12-31T00:00:00=0, 2012-03-31T00:00:00=5, 2012-12-31T00:00:00=0, 2013-03-31T00:00:00=5, 2012-10-31T00:00:00=0, 2013-12-31T00:00:00=0, 2013-10-31T00:00:00=0, 2011-06-30T00:00:00=0, 2011-10-31T00:00:00=0, 2012-06-30T00:00:00=0, 2013-06-30T00:00:00=0),\n",
529 |        " Row(paid_by_another_entity_fiscal_year_end_date=u'N', 2011-03-31T00:00:00=2376, 2011-12-31T00:00:00=35169, 2012-03-31T00:00:00=2348, 2012-12-31T00:00:00=3548, 2013-03-31T00:00:00=2272, 2012-10-31T00:00:00=183, 2013-12-31T00:00:00=3602, 2013-10-31T00:00:00=208, 2011-06-30T00:00:00=24, 2011-10-31T00:00:00=202, 2012-06-30T00:00:00=24, 2013-06-30T00:00:00=24),\n",
530 |        " Row(paid_by_another_entity_fiscal_year_end_date=u'null', 2011-03-31T00:00:00=1, 2011-12-31T00:00:00=0, 2012-03-31T00:00:00=1, 2012-12-31T00:00:00=0, 2013-03-31T00:00:00=1, 2012-10-31T00:00:00=0, 2013-12-31T00:00:00=0, 2013-10-31T00:00:00=0, 2011-06-30T00:00:00=0, 2011-10-31T00:00:00=0, 2012-06-30T00:00:00=0, 2013-06-30T00:00:00=0)]"
531 |       ]
532 |      },
533 |      "execution_count": 27,
534 |      "metadata": {},
535 |      "output_type": "execute_result"
536 |     }
537 |    ],
538 |    "source": [
539 |     "sf(df_cast).crosstab(\"paid_by_another_entity\", \"fiscal_year_end_date\").collect()"
540 |    ]
541 |   },
542 |   {
543 |    "cell_type": "markdown",
544 |    "metadata": {},
545 |    "source": [
546 |     "#Learn More\n",
547 |     "* [Databricks Spark SQL blog](https://databricks.com/blog/category/sparksql)\n",
548 |     "* [Spark SQL API](http://spark.apache.org/docs/latest/api/python/pyspark.sql.html)"
549 |    ]
550 |   },
551 |   {
552 |    "cell_type": "markdown",
553 |    "metadata": {},
554 |    "source": [
555 |     "Well I hope this helped you get started with Spark locally and showcased some of the features of data frames. As always, please feel free to email me with questions / comments."
556 |    ]
557 |   }
558 |  ],
559 |  "metadata": {
560 |   "kernelspec": {
561 |    "display_name": "Python 2",
562 |    "language": "python",
563 |    "name": "python2"
564 |   },
565 |   "language_info": {
566 |    "codemirror_mode": {
567 |     "name": "ipython",
568 |     "version": 2
569 |    },
570 |    "file_extension": ".py",
571 |    "mimetype": "text/x-python",
572 |    "name": "python",
573 |    "nbconvert_exporter": "python",
574 |    "pygments_lexer": "ipython2",
575 |    "version": "2.7.10"
576 |   }
577 |  },
578 |  "nbformat": 4,
579 |  "nbformat_minor": 0
580 | }
581 | 


--------------------------------------------------------------------------------
/LSTM_Debate_Model.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 |       "Using Theano backend.\n",
 15 |       "/Users/tylerfolkman/anaconda/lib/python2.7/site-packages/theano/tensor/signal/downsample.py:5: UserWarning: downsample module has been moved to the pool module.\n",
 16 |       "  warnings.warn(\"downsample module has been moved to the pool module.\")\n"
 17 |      ]
 18 |     }
 19 |    ],
 20 |    "source": [
 21 |     "'''\n",
 22 |     "Most of this code is borrowed from: https://github.com/fchollet/keras/blob/master/examples/lstm_text_generation.py\n",
 23 |     "'''\n",
 24 |     "\n",
 25 |     "import theano\n",
 26 |     "from __future__ import print_function\n",
 27 |     "from keras.models import Sequential, model_from_json\n",
 28 |     "from keras.layers.core import Dense, Activation, Dropout\n",
 29 |     "from keras.layers.recurrent import LSTM\n",
 30 |     "from keras.datasets.data_utils import get_file\n",
 31 |     "import numpy as np\n",
 32 |     "import random\n",
 33 |     "import sys"
 34 |    ]
 35 |   },
 36 |   {
 37 |    "cell_type": "code",
 38 |    "execution_count": 2,
 39 |    "metadata": {
 40 |     "collapsed": false
 41 |    },
 42 |    "outputs": [
 43 |     {
 44 |      "name": "stdout",
 45 |      "output_type": "stream",
 46 |      "text": [
 47 |       "131792\n"
 48 |      ]
 49 |     }
 50 |    ],
 51 |    "source": [
 52 |     "import json\n",
 53 |     "\n",
 54 |     "with open(\"candidate_words_dict.json\") as f:\n",
 55 |     "    candidate_word_dict = json.load(f)\n",
 56 |     "text = []\n",
 57 |     "start_indexes = []\n",
 58 |     "i = 0\n",
 59 |     "for v in candidate_word_dict.values():\n",
 60 |     "    text.append(\"<START>\")\n",
 61 |     "    start_indexes.append(i)\n",
 62 |     "    i+=1\n",
 63 |     "    for word in v:\n",
 64 |     "        if word[-1] == '.':\n",
 65 |     "            text.append(word[:-1].lower())\n",
 66 |     "            text.append(\"<END>\")\n",
 67 |     "            text.append(\"<START>\")\n",
 68 |     "            i+=3\n",
 69 |     "            start_indexes.append(i-1)\n",
 70 |     "        else:\n",
 71 |     "            text.append(word.lower())\n",
 72 |     "            i+=1\n",
 73 |     "print (len(text))"
 74 |    ]
 75 |   },
 76 |   {
 77 |    "cell_type": "code",
 78 |    "execution_count": 3,
 79 |    "metadata": {
 80 |     "collapsed": false
 81 |    },
 82 |    "outputs": [
 83 |     {
 84 |      "name": "stdout",
 85 |      "output_type": "stream",
 86 |      "text": [
 87 |       "total chars: 9198\n"
 88 |      ]
 89 |     }
 90 |    ],
 91 |    "source": [
 92 |     "## Get unique characters and create dictionaries from character to index and index to character\n",
 93 |     "\n",
 94 |     "chars = set(text)\n",
 95 |     "print('total chars:', len(chars))\n",
 96 |     "char_indices = dict((c, i) for i, c in enumerate(chars))\n",
 97 |     "indices_char = dict((i, c) for i, c in enumerate(chars))"
 98 |    ]
 99 |   },
100 |   {
101 |    "cell_type": "code",
102 |    "execution_count": 4,
103 |    "metadata": {
104 |     "collapsed": false
105 |    },
106 |    "outputs": [
107 |     {
108 |      "name": "stdout",
109 |      "output_type": "stream",
110 |      "text": [
111 |       "nb sequences: 43929\n"
112 |      ]
113 |     }
114 |    ],
115 |    "source": [
116 |     "maxlen = 5\n",
117 |     "step = 3\n",
118 |     "sentences = []\n",
119 |     "next_chars = []\n",
120 |     "## i goes to len(text) - maxlen because grabbing in those sets\n",
121 |     "for i in range(0, len(text) - maxlen, step):\n",
122 |     "    # get the first set of max len, next time around move start forward step and get max len chars\n",
123 |     "    sentences.append(text[i: i + maxlen])\n",
124 |     "    # character immediately following the sentences set\n",
125 |     "    next_chars.append(text[i + maxlen])\n",
126 |     "print('nb sequences:', len(sentences))"
127 |    ]
128 |   },
129 |   {
130 |    "cell_type": "code",
131 |    "execution_count": 5,
132 |    "metadata": {
133 |     "collapsed": false
134 |    },
135 |    "outputs": [
136 |     {
137 |      "name": "stdout",
138 |      "output_type": "stream",
139 |      "text": [
140 |       "Vectorization...\n"
141 |      ]
142 |     }
143 |    ],
144 |    "source": [
145 |     "print('Vectorization...')\n",
146 |     "# row for each sentence, column for each word in sentence, depth for possible characters\n",
147 |     "X = np.zeros((len(sentences), maxlen, len(chars)), dtype=np.bool)\n",
148 |     "# row for each sentence prediction, column for each possible character\n",
149 |     "y = np.zeros((len(next_chars), len(chars)), dtype=np.bool)\n",
150 |     "for i, sentence in enumerate(sentences):\n",
151 |     "    for t, char in enumerate(sentence):\n",
152 |     "        # char indicies return index for a given character\n",
153 |     "        X[i, t, char_indices[char]] = 1\n",
154 |     "    y[i, char_indices[next_chars[i]]] = 1"
155 |    ]
156 |   },
157 |   {
158 |    "cell_type": "code",
159 |    "execution_count": 7,
160 |    "metadata": {
161 |     "collapsed": false
162 |    },
163 |    "outputs": [
164 |     {
165 |      "name": "stdout",
166 |      "output_type": "stream",
167 |      "text": [
168 |       "Build model...\n"
169 |      ]
170 |     }
171 |    ],
172 |    "source": [
173 |     "# build the model: 2 stacked LSTM\n",
174 |     "print('Build model...')\n",
175 |     "\n",
176 |     "# once build model and have some weights, can re-use\n",
177 |     "\n",
178 |     "#model = model_from_json(open('candidate_model_architecture.json').read())\n",
179 |     "#model.load_weights('candidate_lstm_weights.h5')\n",
180 |     "\n",
181 |     "model = Sequential()\n",
182 |     "\n",
183 |     "model.add(LSTM(512, return_sequences=True, input_shape=(maxlen, len(chars))))\n",
184 |     "model.add(Dropout(0.2))\n",
185 |     "model.add(LSTM(512, return_sequences=False))\n",
186 |     "model.add(Dropout(0.2))\n",
187 |     "model.add(Dense(len(chars)))\n",
188 |     "model.add(Activation('softmax'))\n",
189 |     "\n",
190 |     "model.compile(loss='categorical_crossentropy', optimizer='rmsprop')"
191 |    ]
192 |   },
193 |   {
194 |    "cell_type": "code",
195 |    "execution_count": 35,
196 |    "metadata": {
197 |     "collapsed": true
198 |    },
199 |    "outputs": [],
200 |    "source": [
201 |     "json_string = model.to_json()\n",
202 |     "open('candidate_model_architecture.json', 'w').write(json_string)"
203 |    ]
204 |   },
205 |   {
206 |    "cell_type": "code",
207 |    "execution_count": 9,
208 |    "metadata": {
209 |     "collapsed": true
210 |    },
211 |    "outputs": [],
212 |    "source": [
213 |     "def sample(a, temperature=1.0):\n",
214 |     "    # helper function to sample an index from a probability array\n",
215 |     "    # take log of probabilities\n",
216 |     "    a = np.log(a) / temperature\n",
217 |     "    # convert back to probabilities\n",
218 |     "    a = np.exp(a) / np.sum(np.exp(a))\n",
219 |     "    # Sample one given the probabilities and get that index\n",
220 |     "    return np.argmax(np.random.multinomial(1, a, 1))"
221 |    ]
222 |   },
223 |   {
224 |    "cell_type": "code",
225 |    "execution_count": 10,
226 |    "metadata": {
227 |     "collapsed": false
228 |    },
229 |    "outputs": [
230 |     {
231 |      "name": "stdout",
232 |      "output_type": "stream",
233 |      "text": [
234 |       "\n",
235 |       "--------------------------------------------------\n",
236 |       "Iteration 1\n",
237 |       "Epoch 1/1\n",
238 |       "43929/43929 [==============================] - 605s - loss: 4.9875   \n",
239 |       "\n",
240 |       "--------------------------------------------------\n",
241 |       "Iteration 2\n",
242 |       "Epoch 1/1\n",
243 |       "43929/43929 [==============================] - 2801s - loss: 4.9129  \n",
244 |       "\n",
245 |       "--------------------------------------------------\n",
246 |       "Iteration 3\n",
247 |       "Epoch 1/1\n",
248 |       "43929/43929 [==============================] - 644s - loss: 4.8380   \n"
249 |      ]
250 |     }
251 |    ],
252 |    "source": [
253 |     "# train the model, output generated text after each iteration\n",
254 |     "for iteration in range(1, 4):\n",
255 |     "    print()\n",
256 |     "    print('-' * 50)\n",
257 |     "    print('Iteration', iteration)\n",
258 |     "    model.fit(X, y, batch_size=64, nb_epoch=1)"
259 |    ]
260 |   },
261 |   {
262 |    "cell_type": "code",
263 |    "execution_count": 11,
264 |    "metadata": {
265 |     "collapsed": false
266 |    },
267 |    "outputs": [],
268 |    "source": [
269 |     "model.save_weights('candidate_lstm_weights_v2.h5', overwrite=True)"
270 |    ]
271 |   },
272 |   {
273 |    "cell_type": "code",
274 |    "execution_count": null,
275 |    "metadata": {
276 |     "collapsed": false
277 |    },
278 |    "outputs": [],
279 |    "source": [
280 |     "start_index = random.choice(start_indexes)\n",
281 |     "\n",
282 |     "for diversity in [.1, .5, .75, .9, 1.0]:\n",
283 |     "    print()\n",
284 |     "    print('----- diversity:', diversity)\n",
285 |     "\n",
286 |     "    generated = []\n",
287 |     "    # get a random sentence of correct length for model\n",
288 |     "    sentence = text[start_index: start_index + maxlen]\n",
289 |     "    generated.extend(sentence)\n",
290 |     "    print('----- Generating with seed: {}\\n'.format(u\" \".join(sentence).encode('utf-8').strip()))\n",
291 |     "\n",
292 |     "    next_char = \"\"\n",
293 |     "    while next_char != \"<END>\":\n",
294 |     "        # convert the sentence into the x format necessary for model\n",
295 |     "        # meaning # sentences, # characters in sentence, # vocab\n",
296 |     "            x = np.zeros((1, maxlen, len(chars)))\n",
297 |     "            for t, char in enumerate(sentence):\n",
298 |     "                x[0, t, char_indices[char]] = 1.\n",
299 |     "\n",
300 |     "        preds = model.predict(x, verbose=0)[0]\n",
301 |     "        next_index = sample(preds, diversity)\n",
302 |     "        next_char = indices_char[next_index]\n",
303 |     "\n",
304 |     "        # generated keeps track of text im generating\n",
305 |     "        generated.append(next_char)\n",
306 |     "        # sentence shifts down by 1 for next loop\n",
307 |     "        sentence = sentence[1:]\n",
308 |     "        sentence.append(next_char)\n",
309 |     "\n",
310 |     "    print(\"Resulting Setence: {}\".format(u\" \".join(generated).encode('utf-8').strip()))"
311 |    ]
312 |   }
313 |  ],
314 |  "metadata": {
315 |   "kernelspec": {
316 |    "display_name": "Python 2",
317 |    "language": "python",
318 |    "name": "python2"
319 |   },
320 |   "language_info": {
321 |    "codemirror_mode": {
322 |     "name": "ipython",
323 |     "version": 2
324 |    },
325 |    "file_extension": ".py",
326 |    "mimetype": "text/x-python",
327 |    "name": "python",
328 |    "nbconvert_exporter": "python",
329 |    "pygments_lexer": "ipython2",
330 |    "version": "2.7.11"
331 |   }
332 |  },
333 |  "nbformat": 4,
334 |  "nbformat_minor": 0
335 | }
336 | 


--------------------------------------------------------------------------------
/NFL_Draft_Analyze.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 34,
  6 |    "metadata": {
  7 |     "collapsed": true
  8 |    },
  9 |    "outputs": [],
 10 |    "source": [
 11 |     "import pandas as pd\n",
 12 |     "import pickle\n",
 13 |     "import seaborn as sns\n",
 14 |     "import matplotlib.pyplot as plt\n",
 15 |     "%matplotlib inline"
 16 |    ]
 17 |   },
 18 |   {
 19 |    "cell_type": "code",
 20 |    "execution_count": 24,
 21 |    "metadata": {
 22 |     "collapsed": false
 23 |    },
 24 |    "outputs": [],
 25 |    "source": [
 26 |     "with open(\"../data/2015_draft\",'r') as f:\n",
 27 |     "    player_data = pickle.load(f)\n",
 28 |     "df = pd.DataFrame(player_data[1:], columns=player_data[0])"
 29 |    ]
 30 |   },
 31 |   {
 32 |    "cell_type": "code",
 33 |    "execution_count": 25,
 34 |    "metadata": {
 35 |     "collapsed": true
 36 |    },
 37 |    "outputs": [],
 38 |    "source": [
 39 |     "df = df.convert_objects(convert_numeric=True)"
 40 |    ]
 41 |   },
 42 |   {
 43 |    "cell_type": "code",
 44 |    "execution_count": 26,
 45 |    "metadata": {
 46 |     "collapsed": false
 47 |    },
 48 |    "outputs": [
 49 |     {
 50 |      "data": {
 51 |       "text/html": [
 52 |        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
 53 |        "<table border=\"1\" class=\"dataframe\">\n",
 54 |        "  <thead>\n",
 55 |        "    <tr style=\"text-align: right;\">\n",
 56 |        "      <th></th>\n",
 57 |        "      <th>Rnd</th>\n",
 58 |        "      <th>Pick</th>\n",
 59 |        "      <th>Tm</th>\n",
 60 |        "      <th>Name</th>\n",
 61 |        "      <th>Pos</th>\n",
 62 |        "      <th>Age</th>\n",
 63 |        "      <th>To</th>\n",
 64 |        "      <th>AP1</th>\n",
 65 |        "      <th>PB</th>\n",
 66 |        "      <th>St</th>\n",
 67 |        "      <th>...</th>\n",
 68 |        "      <th>TD</th>\n",
 69 |        "      <th>Rec</th>\n",
 70 |        "      <th>Yds</th>\n",
 71 |        "      <th>TD</th>\n",
 72 |        "      <th>Tkl</th>\n",
 73 |        "      <th>Int</th>\n",
 74 |        "      <th>Sk</th>\n",
 75 |        "      <th>College/Univ</th>\n",
 76 |        "      <th>College_Stats</th>\n",
 77 |        "      <th>year</th>\n",
 78 |        "    </tr>\n",
 79 |        "  </thead>\n",
 80 |        "  <tbody>\n",
 81 |        "    <tr>\n",
 82 |        "      <th>0</th>\n",
 83 |        "      <td>1</td>\n",
 84 |        "      <td>1</td>\n",
 85 |        "      <td>CLE</td>\n",
 86 |        "      <td>Courtney Brown</td>\n",
 87 |        "      <td>DE</td>\n",
 88 |        "      <td>22</td>\n",
 89 |        "      <td>2005</td>\n",
 90 |        "      <td>0</td>\n",
 91 |        "      <td>0</td>\n",
 92 |        "      <td>4</td>\n",
 93 |        "      <td>...</td>\n",
 94 |        "      <td>NaN</td>\n",
 95 |        "      <td>NaN</td>\n",
 96 |        "      <td>NaN</td>\n",
 97 |        "      <td>NaN</td>\n",
 98 |        "      <td>155</td>\n",
 99 |        "      <td>NaN</td>\n",
100 |        "      <td>19.0</td>\n",
101 |        "      <td>Penn St.</td>\n",
102 |        "      <td>College Stats</td>\n",
103 |        "      <td>2000</td>\n",
104 |        "    </tr>\n",
105 |        "    <tr>\n",
106 |        "      <th>1</th>\n",
107 |        "      <td>1</td>\n",
108 |        "      <td>2</td>\n",
109 |        "      <td>WAS</td>\n",
110 |        "      <td>LaVar Arrington</td>\n",
111 |        "      <td>LB</td>\n",
112 |        "      <td>22</td>\n",
113 |        "      <td>2006</td>\n",
114 |        "      <td>0</td>\n",
115 |        "      <td>3</td>\n",
116 |        "      <td>5</td>\n",
117 |        "      <td>...</td>\n",
118 |        "      <td>NaN</td>\n",
119 |        "      <td>NaN</td>\n",
120 |        "      <td>NaN</td>\n",
121 |        "      <td>NaN</td>\n",
122 |        "      <td>331</td>\n",
123 |        "      <td>3</td>\n",
124 |        "      <td>23.5</td>\n",
125 |        "      <td>Penn St.</td>\n",
126 |        "      <td>College Stats</td>\n",
127 |        "      <td>2000</td>\n",
128 |        "    </tr>\n",
129 |        "    <tr>\n",
130 |        "      <th>2</th>\n",
131 |        "      <td>1</td>\n",
132 |        "      <td>3</td>\n",
133 |        "      <td>WAS</td>\n",
134 |        "      <td>Chris Samuels</td>\n",
135 |        "      <td>T</td>\n",
136 |        "      <td>23</td>\n",
137 |        "      <td>2009</td>\n",
138 |        "      <td>0</td>\n",
139 |        "      <td>6</td>\n",
140 |        "      <td>9</td>\n",
141 |        "      <td>...</td>\n",
142 |        "      <td>NaN</td>\n",
143 |        "      <td>NaN</td>\n",
144 |        "      <td>NaN</td>\n",
145 |        "      <td>NaN</td>\n",
146 |        "      <td>1</td>\n",
147 |        "      <td>NaN</td>\n",
148 |        "      <td>NaN</td>\n",
149 |        "      <td>Alabama</td>\n",
150 |        "      <td>College Stats</td>\n",
151 |        "      <td>2000</td>\n",
152 |        "    </tr>\n",
153 |        "    <tr>\n",
154 |        "      <th>3</th>\n",
155 |        "      <td>1</td>\n",
156 |        "      <td>4</td>\n",
157 |        "      <td>CIN</td>\n",
158 |        "      <td>Peter Warrick</td>\n",
159 |        "      <td>WR</td>\n",
160 |        "      <td>23</td>\n",
161 |        "      <td>2005</td>\n",
162 |        "      <td>0</td>\n",
163 |        "      <td>0</td>\n",
164 |        "      <td>4</td>\n",
165 |        "      <td>...</td>\n",
166 |        "      <td>2</td>\n",
167 |        "      <td>275</td>\n",
168 |        "      <td>2991</td>\n",
169 |        "      <td>18</td>\n",
170 |        "      <td>12</td>\n",
171 |        "      <td>NaN</td>\n",
172 |        "      <td>NaN</td>\n",
173 |        "      <td>Florida St.</td>\n",
174 |        "      <td>College Stats</td>\n",
175 |        "      <td>2000</td>\n",
176 |        "    </tr>\n",
177 |        "    <tr>\n",
178 |        "      <th>4</th>\n",
179 |        "      <td>1</td>\n",
180 |        "      <td>5</td>\n",
181 |        "      <td>BAL</td>\n",
182 |        "      <td>Jamal Lewis</td>\n",
183 |        "      <td>RB</td>\n",
184 |        "      <td>21</td>\n",
185 |        "      <td>2009</td>\n",
186 |        "      <td>1</td>\n",
187 |        "      <td>1</td>\n",
188 |        "      <td>9</td>\n",
189 |        "      <td>...</td>\n",
190 |        "      <td>58</td>\n",
191 |        "      <td>221</td>\n",
192 |        "      <td>1879</td>\n",
193 |        "      <td>4</td>\n",
194 |        "      <td>8</td>\n",
195 |        "      <td>NaN</td>\n",
196 |        "      <td>NaN</td>\n",
197 |        "      <td>Tennessee</td>\n",
198 |        "      <td>College Stats</td>\n",
199 |        "      <td>2000</td>\n",
200 |        "    </tr>\n",
201 |        "  </tbody>\n",
202 |        "</table>\n",
203 |        "<p>5 rows × 30 columns</p>\n",
204 |        "</div>"
205 |       ],
206 |       "text/plain": [
207 |        "   Rnd  Pick   Tm             Name Pos  Age    To  AP1  PB  St  ...   TD  Rec  \\\n",
208 |        "0    1     1  CLE   Courtney Brown  DE   22  2005    0   0   4  ...  NaN  NaN   \n",
209 |        "1    1     2  WAS  LaVar Arrington  LB   22  2006    0   3   5  ...  NaN  NaN   \n",
210 |        "2    1     3  WAS    Chris Samuels   T   23  2009    0   6   9  ...  NaN  NaN   \n",
211 |        "3    1     4  CIN    Peter Warrick  WR   23  2005    0   0   4  ...    2  275   \n",
212 |        "4    1     5  BAL      Jamal Lewis  RB   21  2009    1   1   9  ...   58  221   \n",
213 |        "\n",
214 |        "    Yds  TD  Tkl  Int    Sk  College/Univ  College_Stats  year  \n",
215 |        "0   NaN NaN  155  NaN  19.0      Penn St.  College Stats  2000  \n",
216 |        "1   NaN NaN  331    3  23.5      Penn St.  College Stats  2000  \n",
217 |        "2   NaN NaN    1  NaN   NaN       Alabama  College Stats  2000  \n",
218 |        "3  2991  18   12  NaN   NaN   Florida St.  College Stats  2000  \n",
219 |        "4  1879   4    8  NaN   NaN     Tennessee  College Stats  2000  \n",
220 |        "\n",
221 |        "[5 rows x 30 columns]"
222 |       ]
223 |      },
224 |      "execution_count": 26,
225 |      "metadata": {},
226 |      "output_type": "execute_result"
227 |     }
228 |    ],
229 |    "source": [
230 |     "df.head()"
231 |    ]
232 |   },
233 |   {
234 |    "cell_type": "code",
235 |    "execution_count": 27,
236 |    "metadata": {
237 |     "collapsed": false
238 |    },
239 |    "outputs": [
240 |     {
241 |      "data": {
242 |       "text/plain": [
243 |        "<matplotlib.axes._subplots.AxesSubplot at 0x11becc590>"
244 |       ]
245 |      },
246 |      "execution_count": 27,
247 |      "metadata": {},
248 |      "output_type": "execute_result"
249 |     },
250 |     {
251 |      "data": {
252 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAERCAYAAAB8eMxzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQMgN6ZAN+w4AAAAASUVORK5CYII=\n",
253 |       "text/plain": [
254 |        "<matplotlib.figure.Figure at 0x11bec17d0>"
255 |       ]
256 |      },
257 |      "metadata": {},
258 |      "output_type": "display_data"
259 |     }
260 |    ],
261 |    "source": [
262 |     "sns.distplot(df.CarAV.dropna())"
263 |    ]
264 |   },
265 |   {
266 |    "cell_type": "code",
267 |    "execution_count": 31,
268 |    "metadata": {
269 |     "collapsed": false
270 |    },
271 |    "outputs": [
272 |     {
273 |      "data": {
274 |       "text/plain": [
275 |        "(3709, 30)"
276 |       ]
277 |      },
278 |      "execution_count": 31,
279 |      "metadata": {},
280 |      "output_type": "execute_result"
281 |     }
282 |    ],
283 |    "source": [
284 |     "df_clean = df[pd.notnull(df.CarAV)]\n",
285 |     "df_clean.shape"
286 |    ]
287 |   },
288 |   {
289 |    "cell_type": "code",
290 |    "execution_count": 47,
291 |    "metadata": {
292 |     "collapsed": false
293 |    },
294 |    "outputs": [],
295 |    "source": [
296 |     "top_10_pos = df_clean.Pos.value_counts()[:10].index\n",
297 |     "df_pos = df_clean[df_clean.Pos.isin(top_10_pos)]"
298 |    ]
299 |   },
300 |   {
301 |    "cell_type": "code",
302 |    "execution_count": 49,
303 |    "metadata": {
304 |     "collapsed": false
305 |    },
306 |    "outputs": [
307 |     {
308 |      "data": {
309 |       "text/plain": [
310 |        "(3415, 30)"
311 |       ]
312 |      },
313 |      "execution_count": 49,
314 |      "metadata": {},
315 |      "output_type": "execute_result"
316 |     }
317 |    ],
318 |    "source": [
319 |     "df_pos.shape"
320 |    ]
321 |   },
322 |   {
323 |    "cell_type": "code",
324 |    "execution_count": 55,
325 |    "metadata": {
326 |     "collapsed": false
327 |    },
328 |    "outputs": [
329 |     {
330 |      "data": {
331 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAACGgAAADSCAYAAAAcjmT3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYJC/9Sj2FUczQQMAAAAAAAAAYBbmNEdq10BLswAAAABJRU5ErkJggg==\n",
332 |       "text/plain": [
333 |        "<matplotlib.figure.Figure at 0x121342490>"
334 |       ]
335 |      },
336 |      "metadata": {},
337 |      "output_type": "display_data"
338 |     }
339 |    ],
340 |    "source": [
341 |     "g = sns.FacetGrid(df_pos, col=\"Pos\")\n",
342 |     "g.map(plt.hist, \"CarAV\");"
343 |    ]
344 |   },
345 |   {
346 |    "cell_type": "code",
347 |    "execution_count": 56,
348 |    "metadata": {
349 |     "collapsed": false
350 |    },
351 |    "outputs": [
352 |     {
353 |      "data": {
354 |       "text/plain": [
355 |        "Rnd\n",
356 |        "1    34.477228\n",
357 |        "2    22.810379\n",
358 |        "3    15.054511\n",
359 |        "4    12.107527\n",
360 |        "5     8.713178\n",
361 |        "6     7.211650\n",
362 |        "7     6.280069\n",
363 |        "Name: CarAV, dtype: float64"
364 |       ]
365 |      },
366 |      "execution_count": 56,
367 |      "metadata": {},
368 |      "output_type": "execute_result"
369 |     }
370 |    ],
371 |    "source": [
372 |     "df_clean.groupby('Rnd').CarAV.mean()"
373 |    ]
374 |   },
375 |   {
376 |    "cell_type": "code",
377 |    "execution_count": 69,
378 |    "metadata": {
379 |     "collapsed": false
380 |    },
381 |    "outputs": [],
382 |    "source": [
383 |     "pos_car = df_clean.groupby('Pos').CarAV.mean()"
384 |    ]
385 |   },
386 |   {
387 |    "cell_type": "code",
388 |    "execution_count": 72,
389 |    "metadata": {
390 |     "collapsed": false
391 |    },
392 |    "outputs": [
393 |     {
394 |      "data": {
395 |       "text/plain": [
396 |        "Pos\n",
397 |        "NT     22.222222\n",
398 |        "T      21.215328\n",
399 |        "C      20.228916\n",
400 |        "QB     18.553571\n",
401 |        "G      18.355330\n",
402 |        "DE     17.868657\n",
403 |        "LB     16.454139\n",
404 |        "DT     15.182724\n",
405 |        "RB     14.769231\n",
406 |        "DB     13.488732\n",
407 |        "WR     13.396552\n",
408 |        "P      11.321429\n",
409 |        "K      11.125000\n",
410 |        "TE      8.654545\n",
411 |        "FS      4.000000\n",
412 |        "FB      3.075000\n",
413 |        "ILB     2.818182\n",
414 |        "OLB     2.388889\n",
415 |        "SS      2.285714\n",
416 |        "OL      2.194444\n",
417 |        "CB      1.909091\n",
418 |        "LS      1.000000\n",
419 |        "DL      0.000000\n",
420 |        "Name: CarAV, dtype: float64"
421 |       ]
422 |      },
423 |      "execution_count": 72,
424 |      "metadata": {},
425 |      "output_type": "execute_result"
426 |     }
427 |    ],
428 |    "source": [
429 |     "pos_car.order(ascending=False)"
430 |    ]
431 |   },
432 |   {
433 |    "cell_type": "code",
434 |    "execution_count": 81,
435 |    "metadata": {
436 |     "collapsed": false
437 |    },
438 |    "outputs": [
439 |     {
440 |      "data": {
441 |       "text/html": [
442 |        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
443 |        "<table border=\"1\" class=\"dataframe\">\n",
444 |        "  <thead>\n",
445 |        "    <tr style=\"text-align: right;\">\n",
446 |        "      <th></th>\n",
447 |        "      <th>Name</th>\n",
448 |        "      <th>Pos</th>\n",
449 |        "      <th>Rnd</th>\n",
450 |        "      <th>Pick</th>\n",
451 |        "      <th>CarAV</th>\n",
452 |        "    </tr>\n",
453 |        "  </thead>\n",
454 |        "  <tbody>\n",
455 |        "    <tr>\n",
456 |        "      <th>208</th>\n",
457 |        "      <td>Tom Brady</td>\n",
458 |        "      <td>QB</td>\n",
459 |        "      <td>6</td>\n",
460 |        "      <td>199</td>\n",
461 |        "      <td>160</td>\n",
462 |        "    </tr>\n",
463 |        "    <tr>\n",
464 |        "      <th>299</th>\n",
465 |        "      <td>Drew Brees</td>\n",
466 |        "      <td>QB</td>\n",
467 |        "      <td>2</td>\n",
468 |        "      <td>32</td>\n",
469 |        "      <td>147</td>\n",
470 |        "    </tr>\n",
471 |        "    <tr>\n",
472 |        "      <th>270</th>\n",
473 |        "      <td>LaDainian Tomlinson</td>\n",
474 |        "      <td>RB</td>\n",
475 |        "      <td>1</td>\n",
476 |        "      <td>5</td>\n",
477 |        "      <td>128</td>\n",
478 |        "    </tr>\n",
479 |        "    <tr>\n",
480 |        "      <th>522</th>\n",
481 |        "      <td>Julius Peppers</td>\n",
482 |        "      <td>DE</td>\n",
483 |        "      <td>1</td>\n",
484 |        "      <td>2</td>\n",
485 |        "      <td>126</td>\n",
486 |        "    </tr>\n",
487 |        "    <tr>\n",
488 |        "      <th>1071</th>\n",
489 |        "      <td>Philip Rivers</td>\n",
490 |        "      <td>QB</td>\n",
491 |        "      <td>1</td>\n",
492 |        "      <td>4</td>\n",
493 |        "      <td>121</td>\n",
494 |        "    </tr>\n",
495 |        "    <tr>\n",
496 |        "      <th>8</th>\n",
497 |        "      <td>Brian Urlacher</td>\n",
498 |        "      <td>LB</td>\n",
499 |        "      <td>1</td>\n",
500 |        "      <td>9</td>\n",
501 |        "      <td>118</td>\n",
502 |        "    </tr>\n",
503 |        "    <tr>\n",
504 |        "      <th>295</th>\n",
505 |        "      <td>Reggie Wayne</td>\n",
506 |        "      <td>WR</td>\n",
507 |        "      <td>1</td>\n",
508 |        "      <td>30</td>\n",
509 |        "      <td>117</td>\n",
510 |        "    </tr>\n",
511 |        "    <tr>\n",
512 |        "      <th>1355</th>\n",
513 |        "      <td>Aaron Rodgers</td>\n",
514 |        "      <td>QB</td>\n",
515 |        "      <td>1</td>\n",
516 |        "      <td>24</td>\n",
517 |        "      <td>112</td>\n",
518 |        "    </tr>\n",
519 |        "    <tr>\n",
520 |        "      <th>1078</th>\n",
521 |        "      <td>Ben Roethlisberger</td>\n",
522 |        "      <td>QB</td>\n",
523 |        "      <td>1</td>\n",
524 |        "      <td>11</td>\n",
525 |        "      <td>108</td>\n",
526 |        "    </tr>\n",
527 |        "    <tr>\n",
528 |        "      <th>544</th>\n",
529 |        "      <td>Ed Reed</td>\n",
530 |        "      <td>DB</td>\n",
531 |        "      <td>1</td>\n",
532 |        "      <td>24</td>\n",
533 |        "      <td>106</td>\n",
534 |        "    </tr>\n",
535 |        "  </tbody>\n",
536 |        "</table>\n",
537 |        "</div>"
538 |       ],
539 |       "text/plain": [
540 |        "                     Name Pos  Rnd  Pick  CarAV\n",
541 |        "208             Tom Brady  QB    6   199    160\n",
542 |        "299            Drew Brees  QB    2    32    147\n",
543 |        "270   LaDainian Tomlinson  RB    1     5    128\n",
544 |        "522        Julius Peppers  DE    1     2    126\n",
545 |        "1071        Philip Rivers  QB    1     4    121\n",
546 |        "8          Brian Urlacher  LB    1     9    118\n",
547 |        "295          Reggie Wayne  WR    1    30    117\n",
548 |        "1355        Aaron Rodgers  QB    1    24    112\n",
549 |        "1078   Ben Roethlisberger  QB    1    11    108\n",
550 |        "544               Ed Reed  DB    1    24    106"
551 |       ]
552 |      },
553 |      "execution_count": 81,
554 |      "metadata": {},
555 |      "output_type": "execute_result"
556 |     }
557 |    ],
558 |    "source": [
559 |     "df_clean.sort(['CarAV'], ascending=False)[['Name', 'Pos', 'Rnd', 'Pick', 'CarAV']].head(10)"
560 |    ]
561 |   },
562 |   {
563 |    "cell_type": "code",
564 |    "execution_count": 88,
565 |    "metadata": {
566 |     "collapsed": false
567 |    },
568 |    "outputs": [],
569 |    "source": [
570 |     "from sklearn.cross_validation import train_test_split\n",
571 |     "\n",
572 |     "df_prediction = df[['Pos', 'Rnd', 'Pick', 'CarAV']]\n",
573 |     "df_prediction = df_prediction.dropna()\n",
574 |     "features = ['Pos', 'Rnd', 'Pick']\n",
575 |     "X = df_prediction[features]\n",
576 |     "X = pd.get_dummies(X)\n",
577 |     "y = df_prediction['CarAV']\n",
578 |     "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"
579 |    ]
580 |   },
581 |   {
582 |    "cell_type": "code",
583 |    "execution_count": 94,
584 |    "metadata": {
585 |     "collapsed": false
586 |    },
587 |    "outputs": [
588 |     {
589 |      "name": "stdout",
590 |      "output_type": "stream",
591 |      "text": [
592 |       "Test RMSE: 19.0394556784\n"
593 |      ]
594 |     }
595 |    ],
596 |    "source": [
597 |     "from sklearn.ensemble import RandomForestRegressor\n",
598 |     "import numpy as np\n",
599 |     "from sklearn.metrics import mean_squared_error\n",
600 |     "clf = RandomForestRegressor(n_estimators=100, n_jobs=-1)\n",
601 |     "clf.fit(X_train, y_train)\n",
602 |     "pred = clf.predict(X_test)\n",
603 |     "print(\"Test RMSE: {0}\".format(np.sqrt(mean_squared_error(y_test, pred))))"
604 |    ]
605 |   },
606 |   {
607 |    "cell_type": "code",
608 |    "execution_count": 95,
609 |    "metadata": {
610 |     "collapsed": false
611 |    },
612 |    "outputs": [
613 |     {
614 |      "data": {
615 |       "text/plain": [
616 |        "<matplotlib.axes._subplots.AxesSubplot at 0x12148b850>"
617 |       ]
618 |      },
619 |      "execution_count": 95,
620 |      "metadata": {},
621 |      "output_type": "execute_result"
622 |     },
623 |     {
624 |      "data": {
625 |       "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX4AAAERCAYAAAB8eMxzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/x8CKAO+75k4tQAAAABJRU5ErkJggg==\n",
626 |       "text/plain": [
627 |        "<matplotlib.figure.Figure at 0x12137da90>"
628 |       ]
629 |      },
630 |      "metadata": {},
631 |      "output_type": "display_data"
632 |     }
633 |    ],
634 |    "source": [
635 |     "sns.distplot(y_test - pred)"
636 |    ]
637 |   },
638 |   {
639 |    "cell_type": "code",
640 |    "execution_count": 102,
641 |    "metadata": {
642 |     "collapsed": false
643 |    },
644 |    "outputs": [
645 |     {
646 |      "data": {
647 |       "text/plain": [
648 |        "[(u'Pick', 0.74057355030955418),\n",
649 |        " (u'Pos_WR', 0.027251325026239556),\n",
650 |        " (u'Pos_QB', 0.026284907953210087),\n",
651 |        " (u'Rnd', 0.024608358747377927),\n",
652 |        " (u'Pos_LB', 0.023325271548009431),\n",
653 |        " (u'Pos_DB', 0.023244069808998812),\n",
654 |        " (u'Pos_DE', 0.022416329004330184),\n",
655 |        " (u'Pos_RB', 0.020809612266325495),\n",
656 |        " (u'Pos_T', 0.019328648430449551),\n",
657 |        " (u'Pos_DT', 0.016179252959674748),\n",
658 |        " (u'Pos_G', 0.015783498291433298),\n",
659 |        " (u'Pos_C', 0.010605214872418485),\n",
660 |        " (u'Pos_TE', 0.0098016266470714788),\n",
661 |        " (u'Pos_NT', 0.0061282718000366231),\n",
662 |        " (u'Pos_CB', 0.0033879088823145376),\n",
663 |        " (u'Pos_OLB', 0.0022238058553690535),\n",
664 |        " (u'Pos_K', 0.0021160134638889883),\n",
665 |        " (u'Pos_ILB', 0.0013145299160797032),\n",
666 |        " (u'Pos_P', 0.0012959567575557965),\n",
667 |        " (u'Pos_FB', 0.001160939197053856),\n",
668 |        " (u'Pos_OL', 0.00097489972575304262),\n",
669 |        " (u'Pos_SS', 0.00068067392576227556),\n",
670 |        " (u'Pos_FS', 0.00043987713249530035),\n",
671 |        " (u'Pos_LS', 3.2781030621886093e-05),\n",
672 |        " (u'Pos_DL', 3.2676447975885517e-05)]"
673 |       ]
674 |      },
675 |      "execution_count": 102,
676 |      "metadata": {},
677 |      "output_type": "execute_result"
678 |     }
679 |    ],
680 |    "source": [
681 |     "fi = zip(X.columns, clf.feature_importances_)\n",
682 |     "fi = sorted(fi, key=lambda x:x[1], reverse=True)\n",
683 |     "fi"
684 |    ]
685 |   },
686 |   {
687 |    "cell_type": "code",
688 |    "execution_count": null,
689 |    "metadata": {
690 |     "collapsed": true
691 |    },
692 |    "outputs": [],
693 |    "source": []
694 |   }
695 |  ],
696 |  "metadata": {
697 |   "kernelspec": {
698 |    "display_name": "Python 2",
699 |    "language": "python",
700 |    "name": "python2"
701 |   },
702 |   "language_info": {
703 |    "codemirror_mode": {
704 |     "name": "ipython",
705 |     "version": 2
706 |    },
707 |    "file_extension": ".py",
708 |    "mimetype": "text/x-python",
709 |    "name": "python",
710 |    "nbconvert_exporter": "python",
711 |    "pygments_lexer": "ipython2",
712 |    "version": "2.7.11"
713 |   }
714 |  },
715 |  "nbformat": 4,
716 |  "nbformat_minor": 0
717 | }
718 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | This repository stores the code for my blog: http://learningwithdata.com/. 
2 | 
3 | Each post on my blog is a Jupyter Notebook, which is stored here. See the blog for my context around each Notebook.
4 | 


--------------------------------------------------------------------------------
/TalkPay.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "markdown",
  5 |    "metadata": {},
  6 |    "source": [
  7 |     "#TALKPAY"
  8 |    ]
  9 |   },
 10 |   {
 11 |    "cell_type": "code",
 12 |    "execution_count": 137,
 13 |    "metadata": {
 14 |     "collapsed": true
 15 |    },
 16 |    "outputs": [],
 17 |    "source": [
 18 |     "import tweepy\n",
 19 |     "import json\n",
 20 |     "import re\n",
 21 |     "import matplotlib.pyplot as plt\n",
 22 |     "import seaborn as sns\n",
 23 |     "import numpy as np\n",
 24 |     "import pandas as pd\n",
 25 |     "sns.set(style='ticks', palette='Set2')\n",
 26 |     "%matplotlib inline\n",
 27 |     " \n",
 28 |     "# Consumer keys and access tokens, used for OAuth - insert your keys\n",
 29 |     "consumer_key = ''\n",
 30 |     "consumer_secret = ''\n",
 31 |     "access_token = ''\n",
 32 |     "access_token_secret = ''\n",
 33 |     " \n",
 34 |     "# OAuth process, using the keys and tokens\n",
 35 |     "auth = tweepy.OAuthHandler(consumer_key, consumer_secret)\n",
 36 |     "auth.set_access_token(access_token, access_token_secret)\n",
 37 |     " \n",
 38 |     "# Creation of the actual interface, using authentication\n",
 39 |     "api = tweepy.API(auth)"
 40 |    ]
 41 |   },
 42 |   {
 43 |    "cell_type": "code",
 44 |    "execution_count": null,
 45 |    "metadata": {
 46 |     "collapsed": true
 47 |    },
 48 |    "outputs": [],
 49 |    "source": [
 50 |     "c = tweepy.Cursor(api.search, q='talkpay')\n",
 51 |     "c_iter = c.items()\n",
 52 |     "all_text = []\n",
 53 |     "for tweet in c_iter:\n",
 54 |     "    all_text.append(tweet.text)"
 55 |    ]
 56 |   },
 57 |   {
 58 |    "cell_type": "markdown",
 59 |    "metadata": {},
 60 |    "source": [
 61 |     "I do probably the most naive approach to extracting the data I want. I just pull out the digits that fall in between \\$ and k. For example, $125k returns 125. This obviously doesn't get all the data and even gets some data we might not care about. For example, some people put multiple salaries over time, for which we maybe only want the most recent. For the sake of quickness and getting more data, I include everything my naive approach finds."
 62 |    ]
 63 |   },
 64 |   {
 65 |    "cell_type": "code",
 66 |    "execution_count": 152,
 67 |    "metadata": {
 68 |     "collapsed": false
 69 |    },
 70 |    "outputs": [],
 71 |    "source": [
 72 |     "thous_dollars_raw = [re.findall(r'\\$(\\d*)[k|K]', t, re.DOTALL | re.MULTILINE) for t in all_text]\n",
 73 |     "thous_dollars = []\n",
 74 |     "for dollar in thous_dollars_raw:\n",
 75 |     "    if len(dollar) > 0:\n",
 76 |     "        for d in dollar:\n",
 77 |     "            thous_dollars.append(int(d))\n",
 78 |     "thous_dollars = np.array(thous_dollars)"
 79 |    ]
 80 |   },
 81 |   {
 82 |    "cell_type": "code",
 83 |    "execution_count": 154,
 84 |    "metadata": {
 85 |     "collapsed": false
 86 |    },
 87 |    "outputs": [
 88 |     {
 89 |      "data": {
 90 |       "image/png": [
 91 |        "iVBORw0KGgoAAAANSUhEUgAAAmoAAAFXCAYAAAAS3mKRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\n",
 92 |        "AAALEgAACxIB0t1+/AAAIABJREFUeJzt3XucXWV97/HPTiBRcLjZoKOkjKj8PLVSAQVKuYSC1Vgt\n",
 93 |        "qO2xp4hCUaRSwQsapEoGxSoaOEq5VLmlWEorFLRUA1igQL2AKBUB+4soOWgNNtIg4aLJJPv8sdY0\n",
 94 |        "mzEze08ya++1Zz7v1ysv9rr/9iwzfvOs51lPo9lsIkmSpPqZ1esCJEmStGkGNUmSpJoyqEmSJNWU\n",
 95 |        "QU2SJKmmDGqSJEk1ZVCTJEmqKYOapK6LiOsiYveImBMRWa57eURc0MGxSyPiveXnDRGxU9X1tlx7\n",
 96 |        "+4i4qQvXeWtE/FnV15FUfwY1SV0VEVsBu2XmcuC3gTvKTS8GdungFM3yTy/sCLy8C9c5ANimC9eR\n",
 97 |        "VHNb9boASTNHRHwZCGC7iLgLeC7waEScBhwLbB8RFwNvBT4F7AsMAA3grZn5tXHO+2zgX4DzM/P8\n",
 98 |        "iNgAfAI4DNgWODUzr4mIbYELgBcCOwFrgD8BfgHcA+ySmY9GRANI4A2Z+d2WS10KPL2s/Wbg8cz8\n",
 99 |        "UHn9nwCHZubNEXEk8AeZ+caIOBb4M4p/GD8M/HlmZkTMAc4EDgJmA3cBJ5Y1vxY4LCKeKK9zMTC3\n",
100 |        "/DlclJltWx4lTQ+2qEnqmsx8NfBJYElm7gncRhGGPgycBtyWmccC+wHPzsz9MvPFwGXAKeOcdj5w\n",
101 |        "I/DRzDy/Zf1jmfky4H8Dl0TErwGvAv47M387MwP4JkVwerA8x5HlsYcAq8aENICjgSfL2q8pz0f5\n",
102 |        "34coQhbA4cCVEXEw8GbgwMzcq/zuV5f7nAKsy8y9M/OlwErg45l5DfBPwNllIHsf8E/ld3k1cFAZ\n",
103 |        "JCXNALaoSeq2l1KEHCged95Xfv6f8JGZX4+Ih8t+WrsBC4BHxznfl4EfZeYVY9afW57ruxHxXeCg\n",
104 |        "zPzHiHggIt4JvKA872gr3XkUrXAXAG8v/ztWa0D6KrBLRMyjCGpnAEdHxDBFK9nRwHB5na9FxOhx\n",
105 |        "O0bEjsBrKFoQX1GunwP8dBPXuhq4LCL2oWg1PDEznftPmiFsUZPUNeWjz6OAJWV42gW4IyLeQUu/\n",
106 |        "s4j4feBLwAbgC8BfM/7vq+OAZkS8Z8z69S2fZwEjZfC7CHgMuBy4ouW8NwLbRMShwIHA5yf6Lpm5\n",
107 |        "AbiWInDtC1wIDAJ/BHwtM58oz/25zNyzbIXbC9gvM1eX205s2bYvRevfqGZ5nS9RPKr9PLAn8N2I\n",
108 |        "2G2i2iRNHwY1Sd30R8CKzHwJ8EGK/lZ7lo8sR4Cty/0OA67NzM8A3wJeR9GPC57aqgXwdeAtwAcj\n",
109 |        "4sUt698MEBF7UfSLuxX4PWBpZl4KLAf+YPS8ZSvV+RRB7vLMXLuJ+kda6oCiZfD9wN2ZuQ64CfgY\n",
110 |        "cFW5/Qbg/5R92ADeVq4DuB54ZznydRZFGP1oy3XmlPX/HfDGzPwH4ASKlsVOBl1ImgYMapK6aX/g\n",
111 |        "38rPBwK3tGz7OvCiiPhHitBycNlp/8vAV4Chsm9W62O/0Van5cBHgM9FxGjY2zcivgVcAvxxZj4C\n",
112 |        "LAHeHhF3Av9A0Vr3/JbzXUYRgj4zTv0/Ab4dEfeVjy9vAp5T1gdF+NqZoqWNzLyBYsDAVyLiO8Cb\n",
113 |        "KEInZb0rKAYR3Evx+/i95bZlwIkRsQj4MHBkRPw78A3g6sy8dZz6JE0zjWbTrg6Sppdy1OezM/O/\n",
114 |        "JnncHwNHZebvV1OZJE1OZYMJImI2RZ+N3Sn+1Xs88EtgKUW/k3uAE+wUK6kCk/69EhH/CswD3jDl\n",
115 |        "1UjSZqpy1OdrgA2ZeUA5RP0vy/WnZuat5RvID6d49CBJUyYzZ7ff61eOWVBBKZK0RSrro5aZX6QY\n",
116 |        "4g4wBKwG9m7pW7GMje8ckiRJ0hiVvkctM9dHxFLgCIrRXq9o2fwYsP3mnLecgmYX4MeZObKldUqS\n",
117 |        "JNVR5S+8zcyjI+JZFPP5Pa1l0wDwSLvjy5dHLh5n8/MoRk1JkiRNO5U9+oyIoyLiA+XikxQvn7yz\n",
118 |        "7K8GsJDivUYTyszhzGy0/qEIaJIkSdNalS1qVwFLI+IWipdYngT8B3BhORnxfWx8KaQkSZLGqCyo\n",
119 |        "ZeaTwBs3sWlBVdeUJEmaTpyZQJIkqaYMapIkSTVlUJMkSaopg5okSVJNGdQkSZJqyqAmSZJUUwY1\n",
120 |        "SZKkmjKoSZIk1ZRBTZIkqaYMapIkSTVlUJMkSaopg5okSVJNGdQkSZJqaqteF1BHjUZjDjA0iUNW\n",
121 |        "NJvNtRWVI0mSZiiD2qYNLVyyKAcG57Xdcc3KVSw7+cwAlldfliRJmkkMauMYGJzHDvMHe12GJEma\n",
122 |        "weyjJkmSVFMGNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmqKYOaJElSTRnUJEmSasqgJkmS\n",
123 |        "VFMGNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmqKYOaJElSTRnUJEmSasqgJkmSVFMGNUmS\n",
124 |        "pJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmqqa2qOnFEbA1cAuwKzAXOAH4M/DOwvNztgsz8fFU1\n",
125 |        "SJIk9bPKghpwJLAqM4+KiB2B7wCnA2dl5tkVXleSJGlaqDKoXQlcVX6eBawD9gYiIg4Hvg+8KzMf\n",
126 |        "q7AGSZKkvlVZUMvMxwEiYoAitP0F8DTgwsy8KyJOBRYD76uqhrppNBpzgKFJHLKi2WyuragcSZJU\n",
127 |        "c1W2qBER84GrgfMy8+8jYvvM/Hm5+QvAOR2cY5gi0E0HQwuXLMqBwXltd1yzchXLTj4z2NifT5Ik\n",
128 |        "zTBVDiZ4FnAD8I7MvLlcfV1EnJiZ3wQOBe5sd57MHAaGx5x7CHhgKuvtloHBeewwf7DXZUiSpD5Q\n",
129 |        "ZYvaqcD2wGkRcVq57l3A/42IdcBK4LgKry9JktTXquyjdhJw0iY2HVDVNSVJkqYTX3grSZJUUwY1\n",
130 |        "SZKkmjKoSZIk1ZRBTZIkqaYMapIkSTVV6QtvZ4L1IyMAQ41Go5PdhyotRpIkTSsGtS30xKrVLFyy\n",
131 |        "6PpOZht46O7sQkWSJGm6MKhNgU5nG1izclUXqpEkSdOFfdQkSZJqyqAmSZJUUwY1SZKkmjKoSZIk\n",
132 |        "1ZRBTZIkqaYMapIkSTVlUJMkSaopg5okSVJNGdQkSZJqyqAmSZJUUwY1SZKkmjKoSZIk1ZRBTZIk\n",
133 |        "qaYMapIkSTVlUJMkSaqprXpdgOqt0WjMAYYmcciKZrO5tqJyZjzvhyTNLAY1tTO0cMmiHBic13bH\n",
134 |        "NStXsezkMwNYXn1ZM5b3Q5JmEIOa2hoYnMcO8wd7XYZK3g9JmjnsoyZJklRTBjVJkqSaMqhJkiTV\n",
135 |        "lEFNkiSppgxqkiRJNWVQkyRJqimDmiRJUk35HrVpwjfWS5I0/RjUpg/fWC9J0jRjUJtGfGO9JEnT\n",
136 |        "i33UJEmSasqgJkmSVFMGNUmSpJqqrI9aRGwNXALsCswFzgC+BywFNgD3ACdkZrOqGiRJkvpZlS1q\n",
137 |        "RwKrMvMg4FXAecBZwKnlugZweIXXlyRJ6mtVBrUrgdNarrMO2Cszby3XLQMOq/D6kiRJfa2yR5+Z\n",
138 |        "+ThARAxQhLYPAktadnkM2L6q60uSJPW7St+jFhHzgauB8zLzioj4RMvmAeCRDs4xDCyupkJJkqT6\n",
139 |        "qnIwwbOAG4B3ZObN5eq7IuLgzLwFWAjc2O48mTkMDI859xDwwFTWK0mSVDdVtqidSvFo87SIGO2r\n",
140 |        "dhJwTkTMAe4Drqrw+pIkSX2to6AWEUcCvwF8DHh9Zl7W7pjMPIkimI21YDIFSpIkzVRtR31GxJnA\n",
141 |        "q4HXA1sDx0TE2VUXJkmSNNN18nqOVwJHAb/IzNXAKyj6l0mSJKlCnQS19WOW525inSRJkqZYJ0Ht\n",
142 |        "SuDvgZ0i4t3AbcAVlVYlSZKk9oMJMvPjEfEq4EFgPnBaZv5z5ZVJkiTNcOMGtYg4GBidMP1J4NqW\n",
143 |        "bQe1TAUlSZKkCkzUonY6G4PaphwyxbVIkiSpxbhBLTMXtC5HxDOB9ZnZdtonSZIkbbm2fdQi4qXA\n",
144 |        "3wC7ALMi4j7gLZl5f9XFSZIkzWSdjPq8BPiLzHxmZu4ILAEurbYsSZIkdRLUaB3lmZnXAM+orCJJ\n",
145 |        "kiQBnc31eXNEnAJcQPGi2yOB+yJiZ4DM/K8K65MkSZqxOglqb6AY/fn2MetvL9fvNtVFSZIkqbMX\n",
146 |        "3g51oQ5JkiSN0cmoz52Bc4FDy/1vAo7PzJ9WXJskSdKM1slggs8Ad1A84twV+DpwcZVFSZIkqbM+\n",
147 |        "artl5utalj8REW+uqiBJkiQVOmlR2xARvz66EBG7AmurK0mSJEnQWYvah4CvRcQd5fJ+wHHVlSRJ\n",
148 |        "kiTobNTnP0fEXsDLKVrgjvfdaZIkSdVr++gzIuYCbwPeCNwGHB8Rc6ouTJIkaabrpI/aeRRTRu0N\n",
149 |        "jAAvxFGfkiRJleskqO2dmR8A1mbmY8Cbgb2qLUuSJEmdjvpsfdT5a8CGiuqRJElSqZOg9mngX4Bn\n",
150 |        "R8SngW8Bn6q0KkmSJHU06vOyiPgWcAhFsHtNZt5deWWSJEkz3LhBLSJmAydQDB74amae27WqJEmS\n",
151 |        "NOGjz/OBPwSeAD4QEYu7U5IkSZJg4qB2MLAgMxcBvwu8oTslSZIkCSYOak9m5gaAzHwYR3pKkiR1\n",
152 |        "VSejPkc1K6tCkiRJv2KiUZ+7RsQlQKNc/vWW5WZm/mnl1UmSJM1gEwW19/DUVrRbyuUGfdi61mg0\n",
153 |        "5gBDHe7e6X6VWT8yAjDUaDTa7TpqqLJiOrQZNa9oNptrq6tIkqT+Nm5Qy8ylXayjG4YWLlmUA4Pz\n",
154 |        "2u740N3ZhXIm9sSq1Sxcsuj6TuqF/qt5zcpVLDv5zACWV1+ZJEn9qe0Lb6eTgcF57DB/sO1+a1au\n",
155 |        "6kI17XVaL/RnzZIkaWLjDiaIiGd0sxBJkiQ91USjPm8GiIjzu1SLJEmSWkz06HMgIi4HXhURT2Pj\n",
156 |        "6E9w1KckSVLlJgpqvwcsAA6gGPE5OtpzUqM+I2Jf4OOZeUhE7AlcC3y/3HxBZn5+M+qWJEma9iYa\n",
157 |        "9fkgcFlEfAf4HhDAbOCezBzp5OQR8X7gTcBj5aq9gbMz8+wtqlqSJGkG6GRmgq0pXqHwN8ClwIMR\n",
158 |        "sV+H578feD0bH5vuDfx+RNwSERc5YEGSJGl8nbye4xzgjZl5O0AZ0s4B9ml3YGZeHRFDLatuBz6b\n",
159 |        "mXdFxKnAYuB9k65aM8okX1a8dfnfdVO87yhf0itJ6ppOgtq2oyENIDO/UQ4u2BzXZObPy89foAh8\n",
160 |        "E4qIYYpAp5lrUi8r3nbeTkz1vuBLeiVJ3ddJUFsdEUdk5hcAIuJ1wMObeb3rIuLEzPwmcChwZ7sD\n",
161 |        "MnMYGG5dV7bSPbCZNagPTeZlxVXsK0lSL3QS1I4D/jYiLqboa/YDigECkzE6SvR44LyIWAesLM8t\n",
162 |        "SZKkTWgb1DJzObBP2fF/VmY+OpkLZOYKYP/y83coXvchSZKkNjqe6zMzH2u/lyRJkqZKJ6/nkCRJ\n",
163 |        "Ug+0DWoRcXw3CpEkSdJTddKi9s7Kq5AkSdKv6KSP2o8i4iaKl9X+olzXzMwPV1eWJEmSOglq3yj/\n",
164 |        "O/qKjcZ4O0qSJGnqdPJ6juHy1RzPB74LbOMIUEmSpOp1MpjgUODfgS8Cg8CKiHhl1YVJkiTNdJ0M\n",
165 |        "JvgYcCCwOjP/EzgY+GSlVUmSJKmjoDYrM1eOLmTmvWzsryZJkqSKdDrq87UAEbEDcALwYKVVSZIk\n",
166 |        "qaMWteOBI4H5wA+BPXEydUmSpMp1Murzp8AfR8R2wLrMfLL6siRJktQ2qEXEbwBLKV7PQUR8D3hL\n",
167 |        "Zv6g2tIkSZJmtk76qF0IDGfmlwEi4nXAxcCCCutShdaPjAAMNRodvbt4qNJi+sgkf24AK5rN5trq\n",
168 |        "KpIkTXedBLWnj4Y0gMy8JiJOq7AmVeyJVatZuGTR9QOD89ru+9Dd2YWK+sNkfm5rVq5i2clnBrC8\n",
169 |        "+sokSdPVuEEtInaimC7q2xHxbuAiYD3FwIJbu1OeqjIwOI8d5g+23W/NylVdqKZ/dPpzkyRpKkzU\n",
170 |        "ovZtNr4v7VDgxPJzo1x/UoV1SZIkzXjjBrXMHOpiHZIkSRqjk1GfL6J4b9qOLaubmfmnlVUlSZKk\n",
171 |        "jgYTXANcAdzdss4ppCRJkirWSVBbnZkfrrwSSZIkPUUnQW1pRHwUuBEYGV2ZmY78lCRJqlAnQW0B\n",
172 |        "8HJg/zHrD5nyaiRJkvQ/OglqLwN2z0z7pUmSJHXRrA72+S6wR9WFSJIk6ak6aVF7PsXsBA8Bo/MW\n",
173 |        "NjNzt+rKkiRJUidB7XCK2Qha+RhUkiSpYp0OJthUMLtsakuRJElSq06C2iFsDGpbAwdSTMpuUJMk\n",
174 |        "SapQ26CWmUe3LkfETsDnqypIkiRJhU5GfY71ODA0xXVIkiRpjE4mZb+5ZbEB7AZ8qbKKJEmSBHTW\n",
175 |        "R+30ls9N4GeZeW9F9UjqgUajMYfJtZSvaDaba9vvJknaEuMGtYj49fLjDze1LTMfrKwqSd02tHDJ\n",
176 |        "ohwYnNd2xzUrV7Hs5DMDWF59WZI0s03UonYrm34tx3PK42ZXUpGknhgYnMcO8wd7XYYkqcW4QS0z\n",
177 |        "h1qXI+IZwNnA7wFvq7YsSZIkdTTqMyIOo5jzE+AlmfmV6kqSJEkStBlMULainQW8EnibAU2SJKl7\n",
178 |        "JhpMcBhwIXADRSvams25QETsC3w8Mw+JiBcAS4ENwD3ACZnpvKGSJEmbMFGL2g3AOoo+aXdHROu2\n",
179 |        "Zmbu1u7kEfF+4E3AY+Wqs4FTM/PWiLiAYsL3L2xO4ZIkSdPdREGtbRDrwP3A64HPlct7Zeat5edl\n",
180 |        "FCHQoCZJkrQJE436XLGlJ8/MqyNiqGVVo+XzY8D2W3oNSZu2fmQEYKjRaLTbFZwWTpJqqZOZCabS\n",
181 |        "hpbPA8Aj7Q6IiGFgcVUFSdPVE6tWs3DJous7eYntQ3dnFyqSJE1Wt4PaXRFxcGbeAiwEbmx3QGYO\n",
182 |        "A8Ot68pWugcqqE+aVjp9ie2alau6UI0kabK6FdRGR3a+F7gwIuYA9wFXden6kiRJfafyoFb2ddu/\n",
183 |        "/Px9YEHV15QkSZoOOpqZQJIkSd1nUJMkSaopg5okSVJNGdQkSZJqqtuv55hSIyMjsxuNxkCHu29b\n",
184 |        "aTFSi0m+bBZgRbPZXFtdRZKkftTXQe3JWRsOOOwjJ509Z5unt53YfdXyB/r6u6q/TOZls2tWrmLZ\n",
185 |        "yWcGsLz6yiRJ/aSvw0tj9uzZz93rxTvNHWjfWDZrq9ldqEjaqNOXzUqSNB77qEmSJNWUQU2SJKmm\n",
186 |        "DGqSJEk1ZVCTJEmqKYOaJElSTRnUJEmSasqgJkmSVFMGNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1\n",
187 |        "ZVCTJEmqKYOaJElSTRnUJEmSasqgJkmSVFMGNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmq\n",
188 |        "KYOaJElSTRnUJEmSasqgJkmSVFMGNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmqKYOaJElS\n",
189 |        "TRnUJEmSasqgJkmSVFNb9eKiEfFt4Ofl4g8z89he1CFJklRnXQ9qEfE0gMw8pNvXliRJ6ie9aFH7\n",
190 |        "LWCbiLi+vP6pmXl7D+qQJEmqtV4EtceBT2bmxRHxQmBZROyemRt6UIt6ZP3ICMBQo9HoZPehSouR\n",
191 |        "JKmmehHUlgP3A2Tm9yPiYWAQ+M9N7RwRw8DirlWnrnhi1WoWLll0/cDgvLb7PnR3dqEiSZLqpxdB\n",
192 |        "7RhgD+CEiHgOsB2wcrydM3MYGG5dFxFDwAOVVaiuGBicxw7zB9vut2blqi5UI0lS/fQiqF0MXBoR\n",
193 |        "t5bLx/jYU5Ik6Vd1Pahl5ghwVLevK0mS1G984a0kSVJNGdQkSZJqyqAmSZJUUwY1SZKkmjKoSZIk\n",
194 |        "1VRPJmWXpG5qNBpzmNwMFyuazebaisqZUtP5u0kyqEmaGYYWLlmUncyEsWblKpadfGZQzKLSD6bz\n",
195 |        "d5NmPIOapBmh05kw+tF0/m7STGcfNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmqKYOaJElS\n",
196 |        "TRnUJEmSasqgJkmSVFMGNUmSpJoyqEmSJNWUQU2SJKmmDGqSJEk1ZVCTJEmqKYOaJElSTRnUJEmS\n",
197 |        "amqrXhcgSZLUbxqNxhxgqNP9m83m8s25jkFNkiRp8oYWLlmUA4Pz2u64ZuUqgMbmXMSgJkmStBkG\n",
198 |        "Buexw/zBSq9hHzVJkqSaMqhJkiTVlEFNkiSppgxqkiRJNWVQkyRJqimDmiRJUk0Z1CRJkmrKoCZJ\n",
199 |        "klRTBjVJkqSaMqhJkiTVlEFNkiSppgxqkiRJNWVQkyRJqqmtun3BiJgFnA/sAfwSeGtm/qDbdUiS\n",
200 |        "JNVdL1rUjgDmZOb+wCnAWT2oQZIkqfZ6EdR+B7gOIDNvB17WgxokSZJqr+uPPoHtgEdbltdHxKzM\n",
201 |        "3DCJc8wGaIxs2G75tf/6o62fPrftsasf/MncZ/3mC5/9y0fWtD35z1f8hHVrHqdf9q1LHdN53yrP\n",
202 |        "/fh/PczAwMC+z3zmM5/bbt+BgYFd/nv5ip7+LCZTb11M5ufWb99vOn83qc4m+3cvIoaAH2fmyGSu\n",
203 |        "02g2m5tZ4uaJiLOAb2TmleXyjzJz/gT7DwOLu1SeJElSVZ6XmSsmc0AvWtS+CrwWuDIi9gPunmjn\n",
204 |        "zBwGhlvXRcRc4BfAC4D1lVSpKj0APK/XRWizef/6l/euv3n/+tsDwI8ne1AvWtQabBz1CXBMZi7f\n",
205 |        "jPM0M7MxpcWpK7x3/c3717+8d/3N+9ffNvf+db1FLTObwJ91+7qSJEn9xhfeSpIk1ZRBTZIkqab6\n",
206 |        "Oaid3usCtNm8d/3N+9e/vHf9zfvX3zbr/nV9MIEkSZI6088tapIkSdOaQU2SJKmmDGqSJEk1ZVCT\n",
207 |        "JEmqKYOaJElSTfVirs/NFhGz2Dj91C+Bt2bmD3pblcYTEfsCH8/MQyLiBcBSYANwD3BCZjYj4m3A\n",
208 |        "ccAIcEZmfqlnBQuAiNgauATYFZgLnAF8D+9fX4iI2cCFwO5AEzie4vflUrx/fSEidga+BRxKcc+W\n",
209 |        "4r3rCxHxbeDn5eIPgY+xhfev31rUjgDmZOb+wCnAWT2uR+OIiPdT/J/F3HLV2cCpmXkQ0AAOj4hn\n",
210 |        "A+8E9gdeCXwsIub0ol49xZHAqvJevQo4j+LvmvevP7wG2JCZBwAfBP4S71/fKP+h9BngcYp75e/O\n",
211 |        "PhERTwPIzEPKP8cyBfev34La7wDXAWTm7cDLeluOJnA/8HqK/2EC7JWZt5aflwGHAS8HvpqZ6zLz\n",
212 |        "0fKYPbpeqca6Ejit/DwLWIf3r29k5heBt5eLQ8BqYG/vX9/4JHABsLJc9u9e//gtYJuIuD4iboyI\n",
213 |        "/ZiC+9dvQW074NGW5fXl41DVTGZeTdGkO6rR8nkNsD3F/fz5JtarhzLz8cx8LCIGKELbB3nq7wrv\n",
214 |        "X81l5vqIWAp8Grgc//71hYg4mqI1+4ZyVQPvXT95HPhkZr6SosvB5WO2b9b967eQ8ygw0LI8KzM3\n",
215 |        "9KoYTUrrfdoOeIRfvZ8DFP/6V49FxHzgJuCyzLwC71/fycyjgQAuAp7Wssn7V1/HAK+IiJuBlwJ/\n",
216 |        "A8xr2e69q7fllOEsM78PPAw8q2X7Zt2/fgtqXwVeDVA2Kd7d23I0CXdFxMHl54XArcAdwIERMTci\n",
217 |        "tgf+F0VnS/VQRDwLuAF4f2YuLVd7//pERBwVER8oF58E1gN3ev/qLzMPzswFmXkI8O/Am4HrvHd9\n",
218 |        "4xjKvvMR8RyKAHbDlt6/vhr1CVxD8a+Nr5bLx/SyGHVkdDLZ9wIXlh0m7wOuKke+nAPcRvGPhlMz\n",
219 |        "c22P6tRGp1I0w58WEaN91U4CzvH+9YWrgKURcQuwNcW9+w/8+9ePmvi7s59cDFwaEaN90o6haFXb\n",
220 |        "ovvnpOySJEk11W+PPiVJkmYMg5okSVJNGdQkSZJqyqAmSZJUUwY1SZKkmjKoSZIk1VS/vUdNUs1E\n",
221 |        "xLkU8/DOAV5A8a4gKKYvGgKamXl6b6qbWETcA7w6Mx/cgnO8GjgfuC0zj2pZvxQ4BPhvin8UN4Al\n",
222 |        "mXlZm/MtBW4GbgFuzsznbW5tkvqfQU3SFsnMPweIiF2Bf83MPUe3RcTinhXWmal4keQfAh/NzAs3\n",
223 |        "ce4PjQaziHgecFtE/Gdm3timptE/kmY4g5qkqdIYZ/0+5WwizwUuzczTI2IW8CngdykCyecy8xMR\n",
224 |        "sQBYXE6h09q6dA1wBRvnzTs9M68tp2Y5A9gG2JFi2quryuMeAfYGdin3XxoROwKfA3almJfvGeV1\n",
225 |        "9gA+Q/E78RfAMZl5f+uXiIjXAB+haB37IfB24A+Aw4FDI2JDZl483s8kMx+IiE8D7wBujIjdgc+W\n",
226 |        "dT8OnJiZd27q2Ij4TeCcst6dgbMy868iYhjYD5gPnAs8nWLaoQ3AHZl5/KZviaR+YR81SVVqUASL\n",
227 |        "BRSh6X0R8QzgeIrg9hJgH+AN5SPEsa1Io8tHAA9k5suANwEHlOv/HDg2M/cG3gqc1nLsLpl5IPBa\n",
228 |        "YEm57sPAdzLzJcCZFAGnAbyLIvy8HPgrivDzPyJiZ+CvgcMz87co5h0+NzMvAv6JouVsbEjblHuB\n",
229 |        "F5Wf/xb4VHm+dwNXldPMbMqxwEcycx+KcPvRlm1zMvPFwIXAKRQ/572BDeV8g5L6mEFNUpWawLLM\n",
230 |        "XJeZDwM/A3ai6Lu1NDObmfkkcDlw6ATn+BpwRERcQxHSzii3vQnYIyI+CLwH2LblmBvKz/eW14Qi\n",
231 |        "MF4BkJl3UEyE3AS+BJwbERcBa4G/G1PDPhQtVKN92S4cU+94rYmb+i5PRsS2wPMz8wtlLbdT9GWL\n",
232 |        "cY57L7BNRJxCEdK2bdl2e3mOEYqf053AYuC8zPxJh3VJqimDmqSqrW/53KQINaOd60fNonjs2Byz\n",
233 |        "fmugUT6GfBFFoDsQuKPc/m/AyyjCyUd56u+0XwJkZmsrXXPMPiPlPv8I7FWe910UrWetxgaxBp11\n",
234 |        "HRnbQrgHRXAc+/3bnfNKikes9wIfaDm2SfGoFoDMPIKitbIBXBcRB3VQo6QaM6hJ6oWbgLdExKyI\n",
235 |        "2Ab4k3Ldz4DdImJuROxEEcqIiOMp+pldBZwA7Fz2N3shRZ+264BXArPL84/XwvUV4OjynC+hePTa\n",
236 |        "iIi/A/Zm2oOqAAABRklEQVTJzM9SPD7da8xxdwD7lQMmAI4r622ntZ/ZCyn6p12QmWuAH0TE68pt\n",
237 |        "+1H0v7tnnPoPK7/ntRStgpT9/FrP/2sRcR9wT2YupmhRfEkHNUqqMQcTSJpKmxqpuKl+Z58Bdge+\n",
238 |        "Q9Fq9rnM/CJARHyJouVoBXBruf/lwBURcTewjiK0rC4fVd4bET+lGHAwtwx+Y0dNjn5eDFwaEfcC\n",
239 |        "9wP/UW77OHBRRHyIopXt3a0FZ+ZPI+I44JqyH9kKin5jE31vgA9HxLvK7SPAezLzG+W2NwF/HRGn\n",
240 |        "U7SKvT4z10XE6Plav8Mw8G8R8RBwG/A94Hmt+2TmzyLis8A3I+IJ4P8BS8epS1KfaDSbjgCXJEmq\n",
241 |        "Ix99SpIk1ZRBTZIkqaYMapIkSTVlUJMkSaopg5okSVJNGdQkSZJqyqAmSZJUUwY1SZKkmvr/Ld5A\n",
242 |        "xC1zTzYAAAAASUVORK5CYII=\n"
243 |       ],
244 |       "text/plain": [
245 |        "<matplotlib.figure.Figure at 0x10a8a4610>"
246 |       ]
247 |      },
248 |      "metadata": {},
249 |      "output_type": "display_data"
250 |     }
251 |    ],
252 |    "source": [
253 |     "plt.figure(figsize=(10,5))\n",
254 |     "plt.hist(thous_dollars, bins=50)\n",
255 |     "plt.xlabel(\"Thousands of Dollars\")\n",
256 |     "plt.ylabel(\"Number of People\")\n",
257 |     "plt.title(\"#talkpay tweets\")\n",
258 |     "sns.despine()"
259 |    ]
260 |   },
261 |   {
262 |    "cell_type": "code",
263 |    "execution_count": 142,
264 |    "metadata": {
265 |     "collapsed": false
266 |    },
267 |    "outputs": [
268 |     {
269 |      "data": {
270 |       "text/plain": [
271 |        "count    315.000000\n",
272 |        "mean      96.530159\n",
273 |        "std       69.642089\n",
274 |        "min        2.000000\n",
275 |        "25%       42.000000\n",
276 |        "50%       84.000000\n",
277 |        "75%      135.000000\n",
278 |        "max      500.000000\n",
279 |        "dtype: float64"
280 |       ]
281 |      },
282 |      "execution_count": 142,
283 |      "metadata": {},
284 |      "output_type": "execute_result"
285 |     }
286 |    ],
287 |    "source": [
288 |     "pd_series = pd.Series(thous_dollars)\n",
289 |     "pd_series.describe()"
290 |    ]
291 |   },
292 |   {
293 |    "cell_type": "markdown",
294 |    "metadata": {},
295 |    "source": [
296 |     "It looks like we got 315 salaries from our approach - not too bad. Also, the median salary comes out to be \\$84k with the mean being pulled up to \\$96.5k from the few large outliers we have. Virtually all of our data falls below \\$200k and 75 percent of it below \\$135k. \n",
297 |     "\n",
298 |     "So - it is kind of interesting to look at this data. I would be very hesitant to really infer much given the nature of the data and my naive way of extracting it. But still a fun, quick side project :)"
299 |    ]
300 |   }
301 |  ],
302 |  "metadata": {
303 |   "kernelspec": {
304 |    "display_name": "Python 2",
305 |    "language": "python",
306 |    "name": "python2"
307 |   },
308 |   "language_info": {
309 |    "codemirror_mode": {
310 |     "name": "ipython",
311 |     "version": 2
312 |    },
313 |    "file_extension": ".py",
314 |    "mimetype": "text/x-python",
315 |    "name": "python",
316 |    "nbconvert_exporter": "python",
317 |    "pygments_lexer": "ipython2",
318 |    "version": "2.7.9"
319 |   }
320 |  },
321 |  "nbformat": 4,
322 |  "nbformat_minor": 0
323 | }
324 | 


--------------------------------------------------------------------------------
/Top100_OT.png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/tfolkman/learningwithdata/96538dd54a300bb60b8f468b2410f741ff83943e/Top100_OT.png


--------------------------------------------------------------------------------
/us_state_map_makefile:
--------------------------------------------------------------------------------
 1 | build/gz_2010_us_040_00_20m.zip:
 2 | 	mkdir -p $(dir $@)
 3 | 	curl -o $@ http://www2.census.gov/geo/tiger/GENZ2010/$(notdir $@)
 4 | 
 5 | build/gz_2010_us_040_00_20m.shp: build/gz_2010_us_040_00_20m.zip
 6 | 	unzip -od $(dir $@) $<
 7 | 	touch $@
 8 | 
 9 | build/states.json: build/gz_2010_us_040_00_20m.shp state_data.csv
10 | 	node_modules/.bin/topojson \
11 | 		-o $@ \
12 | 		--id-property='STATE,id' \
13 | 		--external-properties=state_data.csv \
14 | 		--properties='name=state' \
15 | 		--properties='rate=+d.properties["value"]' \
16 | 		--properties='rank=+d.properties["rank"]' \
17 | 		--properties='cost=+d.properties["cost"]' \
18 | 		--properties='number=+d.properties["n_schools"]' \
19 | 		--projection='width = 960, height = 600, d3.geo.albersUsa() \
20 | 			.scale(1280) \
21 | 			.translate([width / 2, height / 2])' \
22 | 		--simplify=.5 \
23 | 		--filter=none \
24 | 		-- states=$<
25 | 
26 | 


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