├── .gitattributes
└── pamap2.ipynb


/.gitattributes:
--------------------------------------------------------------------------------
1 | # Auto detect text files and perform LF normalization
2 | * text=auto
3 | 


--------------------------------------------------------------------------------
/pamap2.ipynb:
--------------------------------------------------------------------------------
   1 | {
   2 |  "cells": [
   3 |   {
   4 |    "cell_type": "markdown",
   5 |    "metadata": {},
   6 |    "source": [
   7 |     "# Data Science Research Methods"
   8 |    ]
   9 |   },
  10 |   {
  11 |    "cell_type": "markdown",
  12 |    "metadata": {},
  13 |    "source": []
  14 |   },
  15 |   {
  16 |    "cell_type": "markdown",
  17 |    "metadata": {},
  18 |    "source": [
  19 |     "### Contents\n",
  20 |     "\n",
  21 |     "#### 1.Overview\n",
  22 |     "#### 2.Preparatory Tasks\n",
  23 |     "#### 3.Data Cleaning\n",
  24 |     "#### 4.Data Correlation\n",
  25 |     "#### 5.Exploratory Data Analysis\n",
  26 |     "#### 6.Hypothesis Testing\n",
  27 |     "#### 7.Modelling\n",
  28 |     "#### 7.Decisions\n",
  29 |     "#### 8.Summary\n",
  30 |     "#### 9. References"
  31 |    ]
  32 |   },
  33 |   {
  34 |    "cell_type": "markdown",
  35 |    "metadata": {},
  36 |    "source": [
  37 |     "### Overview"
  38 |    ]
  39 |   },
  40 |   {
  41 |    "cell_type": "markdown",
  42 |    "metadata": {},
  43 |    "source": [
  44 |     "For this assignment, the dataset used is PAMAP2 which is an Activity Monitoring dataset that covers 18 different physical activities which are taken by 9 different subjects, 8 men and 1 woman, taken using 3 inertial measurement units and a heart rate monitor. \n",
  45 |     "\n",
  46 |     "The outcome of this assignment should be to gain insights based on the results of our analysis to determine how active an individual is based on the physical activities taken which are to be used to create hardware of software.\n",
  47 |     "\n",
  48 |     "The three requirements of this assignment are to:\n",
  49 |     "\n",
  50 |     "1. Carry out thoroughly exploratory data analysis and appropriately handle missing or dirty data\n",
  51 |     "\n",
  52 |     "2. Develop and tests at least one hypothesis for a relationship between a single pair of attributes\n",
  53 |     "\n",
  54 |     "3. Develop and test at least one model which uses multiple attributes to make predictions"
  55 |    ]
  56 |   },
  57 |   {
  58 |    "cell_type": "markdown",
  59 |    "metadata": {},
  60 |    "source": [
  61 |     "### Preparatory Tasks"
  62 |    ]
  63 |   },
  64 |   {
  65 |    "cell_type": "code",
  66 |    "execution_count": 1,
  67 |    "metadata": {},
  68 |    "outputs": [],
  69 |    "source": [
  70 |     "from matplotlib import pyplot as plt\n",
  71 |     "%matplotlib inline\n",
  72 |     "import pandas as pd\n",
  73 |     "import numpy as np\n",
  74 |     "import math\n",
  75 |     "from sklearn.model_selection import train_test_split\n"
  76 |    ]
  77 |   },
  78 |   {
  79 |    "cell_type": "markdown",
  80 |    "metadata": {},
  81 |    "source": [
  82 |     "A list of the file names has to be created in order to load all the files and create the dataframe. Moreover, a dictionary that will hold the names as well as numbers of each different activity has to be created in order to be able to understand which activity is being analysed at each phase.\n",
  83 |     "\n",
  84 |     "Lists for each different category of IMU's have to be put together as well in order to have the column names for the dataframe. IMU's that will be used are for chest, ankle and hand.\n",
  85 |     "\n",
  86 |     "Then all the different lists have to be put together to create the collection of the columns."
  87 |    ]
  88 |   },
  89 |   {
  90 |    "cell_type": "code",
  91 |    "execution_count": 2,
  92 |    "metadata": {},
  93 |    "outputs": [
  94 |     {
  95 |      "data": {
  96 |       "text/plain": [
  97 |        "54"
  98 |       ]
  99 |      },
 100 |      "execution_count": 2,
 101 |      "metadata": {},
 102 |      "output_type": "execute_result"
 103 |     }
 104 |    ],
 105 |    "source": [
 106 |     "# Load data\n",
 107 |     "list_of_files = ['PAMAP2_Dataset/Protocol/subject101.dat',\n",
 108 |     "                 'PAMAP2_Dataset/Protocol/subject102.dat',\n",
 109 |     "                 'PAMAP2_Dataset/Protocol/subject103.dat',\n",
 110 |     "                 'PAMAP2_Dataset/Protocol/subject104.dat',\n",
 111 |     "                 'PAMAP2_Dataset/Protocol/subject105.dat',\n",
 112 |     "                 'PAMAP2_Dataset/Protocol/subject106.dat',\n",
 113 |     "                 'PAMAP2_Dataset/Protocol/subject107.dat',\n",
 114 |     "                 'PAMAP2_Dataset/Protocol/subject108.dat',\n",
 115 |     "                 'PAMAP2_Dataset/Protocol/subject109.dat' ]\n",
 116 |     "\n",
 117 |     "subjectID = [1,2,3,4,5,6,7,8,9]\n",
 118 |     "\n",
 119 |     "activityIDdict = {0: 'transient',\n",
 120 |     "              1: 'lying',\n",
 121 |     "              2: 'sitting',\n",
 122 |     "              3: 'standing',\n",
 123 |     "              4: 'walking',\n",
 124 |     "              5: 'running',\n",
 125 |     "              6: 'cycling',\n",
 126 |     "              7: 'Nordic_walking',\n",
 127 |     "              9: 'watching_TV',\n",
 128 |     "              10: 'computer_work',\n",
 129 |     "              11: 'car driving',\n",
 130 |     "              12: 'ascending_stairs',\n",
 131 |     "              13: 'descending_stairs',\n",
 132 |     "              16: 'vacuum_cleaning',\n",
 133 |     "              17: 'ironing',\n",
 134 |     "              18: 'folding_laundry',\n",
 135 |     "              19: 'house_cleaning',\n",
 136 |     "              20: 'playing_soccer',\n",
 137 |     "              24: 'rope_jumping' }\n",
 138 |     "\n",
 139 |     "colNames = [\"timestamp\", \"activityID\",\"heartrate\"]\n",
 140 |     "\n",
 141 |     "IMUhand = ['handTemperature', \n",
 142 |     "           'handAcc16_1', 'handAcc16_2', 'handAcc16_3', \n",
 143 |     "           'handAcc6_1', 'handAcc6_2', 'handAcc6_3', \n",
 144 |     "           'handGyro1', 'handGyro2', 'handGyro3', \n",
 145 |     "           'handMagne1', 'handMagne2', 'handMagne3',\n",
 146 |     "           'handOrientation1', 'handOrientation2', 'handOrientation3', 'handOrientation4']\n",
 147 |     "\n",
 148 |     "IMUchest = ['chestTemperature', \n",
 149 |     "           'chestAcc16_1', 'chestAcc16_2', 'chestAcc16_3', \n",
 150 |     "           'chestAcc6_1', 'chestAcc6_2', 'chestAcc6_3', \n",
 151 |     "           'chestGyro1', 'chestGyro2', 'chestGyro3', \n",
 152 |     "           'chestMagne1', 'chestMagne2', 'chestMagne3',\n",
 153 |     "           'chestOrientation1', 'chestOrientation2', 'chestOrientation3', 'chestOrientation4']\n",
 154 |     "\n",
 155 |     "IMUankle = ['ankleTemperature', \n",
 156 |     "           'ankleAcc16_1', 'ankleAcc16_2', 'ankleAcc16_3', \n",
 157 |     "           'ankleAcc6_1', 'ankleAcc6_2', 'ankleAcc6_3', \n",
 158 |     "           'ankleGyro1', 'ankleGyro2', 'ankleGyro3', \n",
 159 |     "           'ankleMagne1', 'ankleMagne2', 'ankleMagne3',\n",
 160 |     "           'ankleOrientation1', 'ankleOrientation2', 'ankleOrientation3', 'ankleOrientation4']\n",
 161 |     "\n",
 162 |     "columns = colNames + IMUhand + IMUchest + IMUankle  #all columns in one list\n",
 163 |     "\n",
 164 |     "len(columns)\n"
 165 |    ]
 166 |   },
 167 |   {
 168 |    "cell_type": "code",
 169 |    "execution_count": 3,
 170 |    "metadata": {},
 171 |    "outputs": [
 172 |     {
 173 |      "data": {
 174 |       "text/html": [
 175 |        "<div>\n",
 176 |        "<style scoped>\n",
 177 |        "    .dataframe tbody tr th:only-of-type {\n",
 178 |        "        vertical-align: middle;\n",
 179 |        "    }\n",
 180 |        "\n",
 181 |        "    .dataframe tbody tr th {\n",
 182 |        "        vertical-align: top;\n",
 183 |        "    }\n",
 184 |        "\n",
 185 |        "    .dataframe thead th {\n",
 186 |        "        text-align: right;\n",
 187 |        "    }\n",
 188 |        "</style>\n",
 189 |        "<table border=\"1\" class=\"dataframe\">\n",
 190 |        "  <thead>\n",
 191 |        "    <tr style=\"text-align: right;\">\n",
 192 |        "      <th></th>\n",
 193 |        "      <th>timestamp</th>\n",
 194 |        "      <th>activityID</th>\n",
 195 |        "      <th>heartrate</th>\n",
 196 |        "      <th>handTemperature</th>\n",
 197 |        "      <th>handAcc16_1</th>\n",
 198 |        "      <th>handAcc16_2</th>\n",
 199 |        "      <th>handAcc16_3</th>\n",
 200 |        "      <th>handAcc6_1</th>\n",
 201 |        "      <th>handAcc6_2</th>\n",
 202 |        "      <th>handAcc6_3</th>\n",
 203 |        "      <th>...</th>\n",
 204 |        "      <th>ankleGyro2</th>\n",
 205 |        "      <th>ankleGyro3</th>\n",
 206 |        "      <th>ankleMagne1</th>\n",
 207 |        "      <th>ankleMagne2</th>\n",
 208 |        "      <th>ankleMagne3</th>\n",
 209 |        "      <th>ankleOrientation1</th>\n",
 210 |        "      <th>ankleOrientation2</th>\n",
 211 |        "      <th>ankleOrientation3</th>\n",
 212 |        "      <th>ankleOrientation4</th>\n",
 213 |        "      <th>subject_id</th>\n",
 214 |        "    </tr>\n",
 215 |        "  </thead>\n",
 216 |        "  <tbody>\n",
 217 |        "    <tr>\n",
 218 |        "      <th>0</th>\n",
 219 |        "      <td>8.38</td>\n",
 220 |        "      <td>0</td>\n",
 221 |        "      <td>104.0</td>\n",
 222 |        "      <td>30.0</td>\n",
 223 |        "      <td>2.37223</td>\n",
 224 |        "      <td>8.60074</td>\n",
 225 |        "      <td>3.51048</td>\n",
 226 |        "      <td>2.43954</td>\n",
 227 |        "      <td>8.76165</td>\n",
 228 |        "      <td>3.35465</td>\n",
 229 |        "      <td>...</td>\n",
 230 |        "      <td>0.009250</td>\n",
 231 |        "      <td>-0.017580</td>\n",
 232 |        "      <td>-61.1888</td>\n",
 233 |        "      <td>-38.9599</td>\n",
 234 |        "      <td>-58.1438</td>\n",
 235 |        "      <td>1.0</td>\n",
 236 |        "      <td>0.0</td>\n",
 237 |        "      <td>0.0</td>\n",
 238 |        "      <td>0.0</td>\n",
 239 |        "      <td>1</td>\n",
 240 |        "    </tr>\n",
 241 |        "    <tr>\n",
 242 |        "      <th>1</th>\n",
 243 |        "      <td>8.39</td>\n",
 244 |        "      <td>0</td>\n",
 245 |        "      <td>NaN</td>\n",
 246 |        "      <td>30.0</td>\n",
 247 |        "      <td>2.18837</td>\n",
 248 |        "      <td>8.56560</td>\n",
 249 |        "      <td>3.66179</td>\n",
 250 |        "      <td>2.39494</td>\n",
 251 |        "      <td>8.55081</td>\n",
 252 |        "      <td>3.64207</td>\n",
 253 |        "      <td>...</td>\n",
 254 |        "      <td>-0.004638</td>\n",
 255 |        "      <td>0.000368</td>\n",
 256 |        "      <td>-59.8479</td>\n",
 257 |        "      <td>-38.8919</td>\n",
 258 |        "      <td>-58.5253</td>\n",
 259 |        "      <td>1.0</td>\n",
 260 |        "      <td>0.0</td>\n",
 261 |        "      <td>0.0</td>\n",
 262 |        "      <td>0.0</td>\n",
 263 |        "      <td>1</td>\n",
 264 |        "    </tr>\n",
 265 |        "    <tr>\n",
 266 |        "      <th>2</th>\n",
 267 |        "      <td>8.40</td>\n",
 268 |        "      <td>0</td>\n",
 269 |        "      <td>NaN</td>\n",
 270 |        "      <td>30.0</td>\n",
 271 |        "      <td>2.37357</td>\n",
 272 |        "      <td>8.60107</td>\n",
 273 |        "      <td>3.54898</td>\n",
 274 |        "      <td>2.30514</td>\n",
 275 |        "      <td>8.53644</td>\n",
 276 |        "      <td>3.73280</td>\n",
 277 |        "      <td>...</td>\n",
 278 |        "      <td>0.000148</td>\n",
 279 |        "      <td>0.022495</td>\n",
 280 |        "      <td>-60.7361</td>\n",
 281 |        "      <td>-39.4138</td>\n",
 282 |        "      <td>-58.3999</td>\n",
 283 |        "      <td>1.0</td>\n",
 284 |        "      <td>0.0</td>\n",
 285 |        "      <td>0.0</td>\n",
 286 |        "      <td>0.0</td>\n",
 287 |        "      <td>1</td>\n",
 288 |        "    </tr>\n",
 289 |        "    <tr>\n",
 290 |        "      <th>3</th>\n",
 291 |        "      <td>8.41</td>\n",
 292 |        "      <td>0</td>\n",
 293 |        "      <td>NaN</td>\n",
 294 |        "      <td>30.0</td>\n",
 295 |        "      <td>2.07473</td>\n",
 296 |        "      <td>8.52853</td>\n",
 297 |        "      <td>3.66021</td>\n",
 298 |        "      <td>2.33528</td>\n",
 299 |        "      <td>8.53622</td>\n",
 300 |        "      <td>3.73277</td>\n",
 301 |        "      <td>...</td>\n",
 302 |        "      <td>-0.020301</td>\n",
 303 |        "      <td>0.011275</td>\n",
 304 |        "      <td>-60.4091</td>\n",
 305 |        "      <td>-38.7635</td>\n",
 306 |        "      <td>-58.3956</td>\n",
 307 |        "      <td>1.0</td>\n",
 308 |        "      <td>0.0</td>\n",
 309 |        "      <td>0.0</td>\n",
 310 |        "      <td>0.0</td>\n",
 311 |        "      <td>1</td>\n",
 312 |        "    </tr>\n",
 313 |        "    <tr>\n",
 314 |        "      <th>4</th>\n",
 315 |        "      <td>8.42</td>\n",
 316 |        "      <td>0</td>\n",
 317 |        "      <td>NaN</td>\n",
 318 |        "      <td>30.0</td>\n",
 319 |        "      <td>2.22936</td>\n",
 320 |        "      <td>8.83122</td>\n",
 321 |        "      <td>3.70000</td>\n",
 322 |        "      <td>2.23055</td>\n",
 323 |        "      <td>8.59741</td>\n",
 324 |        "      <td>3.76295</td>\n",
 325 |        "      <td>...</td>\n",
 326 |        "      <td>-0.014303</td>\n",
 327 |        "      <td>-0.002823</td>\n",
 328 |        "      <td>-61.5199</td>\n",
 329 |        "      <td>-39.3879</td>\n",
 330 |        "      <td>-58.2694</td>\n",
 331 |        "      <td>1.0</td>\n",
 332 |        "      <td>0.0</td>\n",
 333 |        "      <td>0.0</td>\n",
 334 |        "      <td>0.0</td>\n",
 335 |        "      <td>1</td>\n",
 336 |        "    </tr>\n",
 337 |        "  </tbody>\n",
 338 |        "</table>\n",
 339 |        "<p>5 rows × 55 columns</p>\n",
 340 |        "</div>"
 341 |       ],
 342 |       "text/plain": [
 343 |        "   timestamp  activityID  heartrate  handTemperature  handAcc16_1  \\\n",
 344 |        "0       8.38           0      104.0             30.0      2.37223   \n",
 345 |        "1       8.39           0        NaN             30.0      2.18837   \n",
 346 |        "2       8.40           0        NaN             30.0      2.37357   \n",
 347 |        "3       8.41           0        NaN             30.0      2.07473   \n",
 348 |        "4       8.42           0        NaN             30.0      2.22936   \n",
 349 |        "\n",
 350 |        "   handAcc16_2  handAcc16_3  handAcc6_1  handAcc6_2  handAcc6_3     ...      \\\n",
 351 |        "0      8.60074      3.51048     2.43954     8.76165     3.35465     ...       \n",
 352 |        "1      8.56560      3.66179     2.39494     8.55081     3.64207     ...       \n",
 353 |        "2      8.60107      3.54898     2.30514     8.53644     3.73280     ...       \n",
 354 |        "3      8.52853      3.66021     2.33528     8.53622     3.73277     ...       \n",
 355 |        "4      8.83122      3.70000     2.23055     8.59741     3.76295     ...       \n",
 356 |        "\n",
 357 |        "   ankleGyro2  ankleGyro3  ankleMagne1  ankleMagne2  ankleMagne3  \\\n",
 358 |        "0    0.009250   -0.017580     -61.1888     -38.9599     -58.1438   \n",
 359 |        "1   -0.004638    0.000368     -59.8479     -38.8919     -58.5253   \n",
 360 |        "2    0.000148    0.022495     -60.7361     -39.4138     -58.3999   \n",
 361 |        "3   -0.020301    0.011275     -60.4091     -38.7635     -58.3956   \n",
 362 |        "4   -0.014303   -0.002823     -61.5199     -39.3879     -58.2694   \n",
 363 |        "\n",
 364 |        "   ankleOrientation1  ankleOrientation2  ankleOrientation3  ankleOrientation4  \\\n",
 365 |        "0                1.0                0.0                0.0                0.0   \n",
 366 |        "1                1.0                0.0                0.0                0.0   \n",
 367 |        "2                1.0                0.0                0.0                0.0   \n",
 368 |        "3                1.0                0.0                0.0                0.0   \n",
 369 |        "4                1.0                0.0                0.0                0.0   \n",
 370 |        "\n",
 371 |        "   subject_id  \n",
 372 |        "0           1  \n",
 373 |        "1           1  \n",
 374 |        "2           1  \n",
 375 |        "3           1  \n",
 376 |        "4           1  \n",
 377 |        "\n",
 378 |        "[5 rows x 55 columns]"
 379 |       ]
 380 |      },
 381 |      "execution_count": 3,
 382 |      "metadata": {},
 383 |      "output_type": "execute_result"
 384 |     }
 385 |    ],
 386 |    "source": [
 387 |     "dataCollection = pd.DataFrame()\n",
 388 |     "for file in list_of_files:\n",
 389 |     "    procData = pd.read_table(file, header=None, sep='\\s+')\n",
 390 |     "    procData.columns = columns\n",
 391 |     "    procData['subject_id'] = int(file[-5])\n",
 392 |     "    dataCollection = dataCollection.append(procData, ignore_index=True)\n",
 393 |     "\n",
 394 |     "dataCollection.reset_index(drop=True, inplace=True)\n",
 395 |     "dataCollection.head()"
 396 |    ]
 397 |   },
 398 |   {
 399 |    "cell_type": "markdown",
 400 |    "metadata": {},
 401 |    "source": [
 402 |     "As it can be seen from the sample of the dataframe that we produced, there is some data cleaning required. For instance, activityID 0 must be removed completely from our dataset since this is transient period where the subject was not doing any particular activity as discussed in the readme file that was given. Data cleaning will be discussed thoroughly in the following section."
 403 |    ]
 404 |   },
 405 |   {
 406 |    "cell_type": "markdown",
 407 |    "metadata": {},
 408 |    "source": [
 409 |     "### Data Cleaning"
 410 |    ]
 411 |   },
 412 |   {
 413 |    "cell_type": "markdown",
 414 |    "metadata": {},
 415 |    "source": [
 416 |     "From a look at the **PerformedActivitiesSummary** file which is part of the collection of files given, it can be seen that various data is missing and as the **readme** file comments on, there were some wireless disconnections in data collection therefore the missing data has to be accounted for and made up in a way that our data analysis will not be impacted. More in depth, each different activity has 8 or less subjects which did the activity and each different activity has NaN values for various subjects. Therefore some data filling has be applied. As for the NaN values in our data, it is best to use **interpolate** which is constructing a new data point out of a set of known data points.\n",
 417 |     "\n",
 418 |     "As a guideline for the code written below, interpolation happens after removing 'activity 0' as it has very noisy and would data which would impact the output massively as the values given in heart rate would not be produced from the data point set that is of activities that actually matter. "
 419 |    ]
 420 |   },
 421 |   {
 422 |    "cell_type": "code",
 423 |    "execution_count": 4,
 424 |    "metadata": {},
 425 |    "outputs": [],
 426 |    "source": [
 427 |     "def dataCleaning(dataCollection):\n",
 428 |     "        dataCollection = dataCollection.drop(['handOrientation1', 'handOrientation2', 'handOrientation3', 'handOrientation4',\n",
 429 |     "                                             'chestOrientation1', 'chestOrientation2', 'chestOrientation3', 'chestOrientation4',\n",
 430 |     "                                             'ankleOrientation1', 'ankleOrientation2', 'ankleOrientation3', 'ankleOrientation4'],\n",
 431 |     "                                             axis = 1)  # removal of orientation columns as they are not needed\n",
 432 |     "        dataCollection = dataCollection.drop(dataCollection[dataCollection.activityID == 0].index) #removal of any row of activity 0 as it is transient activity which it is not used\n",
 433 |     "        dataCollection = dataCollection.apply(pd.to_numeric, errors = 'coerse') #removal of non numeric data in cells\n",
 434 |     "        dataCollection = dataCollection.interpolate() #removal of any remaining NaN value cells by constructing new data points in known set of data points\n",
 435 |     "        \n",
 436 |     "        return dataCollection\n"
 437 |    ]
 438 |   },
 439 |   {
 440 |    "cell_type": "code",
 441 |    "execution_count": 5,
 442 |    "metadata": {},
 443 |    "outputs": [],
 444 |    "source": [
 445 |     "dataCol = dataCleaning(dataCollection)"
 446 |    ]
 447 |   },
 448 |   {
 449 |    "cell_type": "code",
 450 |    "execution_count": 6,
 451 |    "metadata": {},
 452 |    "outputs": [
 453 |     {
 454 |      "data": {
 455 |       "text/html": [
 456 |        "<div>\n",
 457 |        "<style scoped>\n",
 458 |        "    .dataframe tbody tr th:only-of-type {\n",
 459 |        "        vertical-align: middle;\n",
 460 |        "    }\n",
 461 |        "\n",
 462 |        "    .dataframe tbody tr th {\n",
 463 |        "        vertical-align: top;\n",
 464 |        "    }\n",
 465 |        "\n",
 466 |        "    .dataframe thead th {\n",
 467 |        "        text-align: right;\n",
 468 |        "    }\n",
 469 |        "</style>\n",
 470 |        "<table border=\"1\" class=\"dataframe\">\n",
 471 |        "  <thead>\n",
 472 |        "    <tr style=\"text-align: right;\">\n",
 473 |        "      <th></th>\n",
 474 |        "      <th>timestamp</th>\n",
 475 |        "      <th>activityID</th>\n",
 476 |        "      <th>heartrate</th>\n",
 477 |        "      <th>handTemperature</th>\n",
 478 |        "      <th>handAcc16_1</th>\n",
 479 |        "      <th>handAcc16_2</th>\n",
 480 |        "      <th>handAcc16_3</th>\n",
 481 |        "      <th>handAcc6_1</th>\n",
 482 |        "      <th>handAcc6_2</th>\n",
 483 |        "      <th>handAcc6_3</th>\n",
 484 |        "      <th>...</th>\n",
 485 |        "      <th>ankleAcc6_1</th>\n",
 486 |        "      <th>ankleAcc6_2</th>\n",
 487 |        "      <th>ankleAcc6_3</th>\n",
 488 |        "      <th>ankleGyro1</th>\n",
 489 |        "      <th>ankleGyro2</th>\n",
 490 |        "      <th>ankleGyro3</th>\n",
 491 |        "      <th>ankleMagne1</th>\n",
 492 |        "      <th>ankleMagne2</th>\n",
 493 |        "      <th>ankleMagne3</th>\n",
 494 |        "      <th>subject_id</th>\n",
 495 |        "    </tr>\n",
 496 |        "  </thead>\n",
 497 |        "  <tbody>\n",
 498 |        "    <tr>\n",
 499 |        "      <th>0</th>\n",
 500 |        "      <td>37.66</td>\n",
 501 |        "      <td>1</td>\n",
 502 |        "      <td>NaN</td>\n",
 503 |        "      <td>30.375</td>\n",
 504 |        "      <td>2.21530</td>\n",
 505 |        "      <td>8.27915</td>\n",
 506 |        "      <td>5.58753</td>\n",
 507 |        "      <td>2.24689</td>\n",
 508 |        "      <td>8.55387</td>\n",
 509 |        "      <td>5.77143</td>\n",
 510 |        "      <td>...</td>\n",
 511 |        "      <td>9.63162</td>\n",
 512 |        "      <td>-1.76757</td>\n",
 513 |        "      <td>0.265761</td>\n",
 514 |        "      <td>0.002908</td>\n",
 515 |        "      <td>-0.027714</td>\n",
 516 |        "      <td>0.001752</td>\n",
 517 |        "      <td>-61.1081</td>\n",
 518 |        "      <td>-36.8636</td>\n",
 519 |        "      <td>-58.3696</td>\n",
 520 |        "      <td>1</td>\n",
 521 |        "    </tr>\n",
 522 |        "    <tr>\n",
 523 |        "      <th>1</th>\n",
 524 |        "      <td>37.67</td>\n",
 525 |        "      <td>1</td>\n",
 526 |        "      <td>NaN</td>\n",
 527 |        "      <td>30.375</td>\n",
 528 |        "      <td>2.29196</td>\n",
 529 |        "      <td>7.67288</td>\n",
 530 |        "      <td>5.74467</td>\n",
 531 |        "      <td>2.27373</td>\n",
 532 |        "      <td>8.14592</td>\n",
 533 |        "      <td>5.78739</td>\n",
 534 |        "      <td>...</td>\n",
 535 |        "      <td>9.58649</td>\n",
 536 |        "      <td>-1.75247</td>\n",
 537 |        "      <td>0.250816</td>\n",
 538 |        "      <td>0.020882</td>\n",
 539 |        "      <td>0.000945</td>\n",
 540 |        "      <td>0.006007</td>\n",
 541 |        "      <td>-60.8916</td>\n",
 542 |        "      <td>-36.3197</td>\n",
 543 |        "      <td>-58.3656</td>\n",
 544 |        "      <td>1</td>\n",
 545 |        "    </tr>\n",
 546 |        "    <tr>\n",
 547 |        "      <th>2</th>\n",
 548 |        "      <td>37.68</td>\n",
 549 |        "      <td>1</td>\n",
 550 |        "      <td>NaN</td>\n",
 551 |        "      <td>30.375</td>\n",
 552 |        "      <td>2.29090</td>\n",
 553 |        "      <td>7.14240</td>\n",
 554 |        "      <td>5.82342</td>\n",
 555 |        "      <td>2.26966</td>\n",
 556 |        "      <td>7.66268</td>\n",
 557 |        "      <td>5.78846</td>\n",
 558 |        "      <td>...</td>\n",
 559 |        "      <td>9.60196</td>\n",
 560 |        "      <td>-1.73721</td>\n",
 561 |        "      <td>0.356632</td>\n",
 562 |        "      <td>-0.035392</td>\n",
 563 |        "      <td>-0.052422</td>\n",
 564 |        "      <td>-0.004882</td>\n",
 565 |        "      <td>-60.3407</td>\n",
 566 |        "      <td>-35.7842</td>\n",
 567 |        "      <td>-58.6119</td>\n",
 568 |        "      <td>1</td>\n",
 569 |        "    </tr>\n",
 570 |        "    <tr>\n",
 571 |        "      <th>3</th>\n",
 572 |        "      <td>37.69</td>\n",
 573 |        "      <td>1</td>\n",
 574 |        "      <td>NaN</td>\n",
 575 |        "      <td>30.375</td>\n",
 576 |        "      <td>2.21800</td>\n",
 577 |        "      <td>7.14365</td>\n",
 578 |        "      <td>5.89930</td>\n",
 579 |        "      <td>2.22177</td>\n",
 580 |        "      <td>7.25535</td>\n",
 581 |        "      <td>5.88000</td>\n",
 582 |        "      <td>...</td>\n",
 583 |        "      <td>9.58674</td>\n",
 584 |        "      <td>-1.78264</td>\n",
 585 |        "      <td>0.311453</td>\n",
 586 |        "      <td>-0.032514</td>\n",
 587 |        "      <td>-0.018844</td>\n",
 588 |        "      <td>0.026950</td>\n",
 589 |        "      <td>-60.7646</td>\n",
 590 |        "      <td>-37.1028</td>\n",
 591 |        "      <td>-57.8799</td>\n",
 592 |        "      <td>1</td>\n",
 593 |        "    </tr>\n",
 594 |        "    <tr>\n",
 595 |        "      <th>4</th>\n",
 596 |        "      <td>37.70</td>\n",
 597 |        "      <td>1</td>\n",
 598 |        "      <td>100.0</td>\n",
 599 |        "      <td>30.375</td>\n",
 600 |        "      <td>2.30106</td>\n",
 601 |        "      <td>7.25857</td>\n",
 602 |        "      <td>6.09259</td>\n",
 603 |        "      <td>2.20720</td>\n",
 604 |        "      <td>7.24042</td>\n",
 605 |        "      <td>5.95555</td>\n",
 606 |        "      <td>...</td>\n",
 607 |        "      <td>9.64677</td>\n",
 608 |        "      <td>-1.75240</td>\n",
 609 |        "      <td>0.295902</td>\n",
 610 |        "      <td>0.001351</td>\n",
 611 |        "      <td>-0.048878</td>\n",
 612 |        "      <td>-0.006328</td>\n",
 613 |        "      <td>-60.2040</td>\n",
 614 |        "      <td>-37.1225</td>\n",
 615 |        "      <td>-57.8847</td>\n",
 616 |        "      <td>1</td>\n",
 617 |        "    </tr>\n",
 618 |        "    <tr>\n",
 619 |        "      <th>5</th>\n",
 620 |        "      <td>37.71</td>\n",
 621 |        "      <td>1</td>\n",
 622 |        "      <td>100.0</td>\n",
 623 |        "      <td>30.375</td>\n",
 624 |        "      <td>2.07165</td>\n",
 625 |        "      <td>7.25965</td>\n",
 626 |        "      <td>6.01218</td>\n",
 627 |        "      <td>2.19238</td>\n",
 628 |        "      <td>7.21038</td>\n",
 629 |        "      <td>6.01604</td>\n",
 630 |        "      <td>...</td>\n",
 631 |        "      <td>9.60177</td>\n",
 632 |        "      <td>-1.75239</td>\n",
 633 |        "      <td>0.311276</td>\n",
 634 |        "      <td>0.003793</td>\n",
 635 |        "      <td>-0.026906</td>\n",
 636 |        "      <td>0.004125</td>\n",
 637 |        "      <td>-61.3257</td>\n",
 638 |        "      <td>-36.9744</td>\n",
 639 |        "      <td>-57.7501</td>\n",
 640 |        "      <td>1</td>\n",
 641 |        "    </tr>\n",
 642 |        "    <tr>\n",
 643 |        "      <th>6</th>\n",
 644 |        "      <td>37.72</td>\n",
 645 |        "      <td>1</td>\n",
 646 |        "      <td>100.0</td>\n",
 647 |        "      <td>30.375</td>\n",
 648 |        "      <td>2.41148</td>\n",
 649 |        "      <td>7.59780</td>\n",
 650 |        "      <td>5.93915</td>\n",
 651 |        "      <td>2.23988</td>\n",
 652 |        "      <td>7.46679</td>\n",
 653 |        "      <td>6.03053</td>\n",
 654 |        "      <td>...</td>\n",
 655 |        "      <td>9.67694</td>\n",
 656 |        "      <td>-1.76748</td>\n",
 657 |        "      <td>0.326060</td>\n",
 658 |        "      <td>0.036814</td>\n",
 659 |        "      <td>-0.032277</td>\n",
 660 |        "      <td>-0.006866</td>\n",
 661 |        "      <td>-61.5520</td>\n",
 662 |        "      <td>-36.9632</td>\n",
 663 |        "      <td>-57.9957</td>\n",
 664 |        "      <td>1</td>\n",
 665 |        "    </tr>\n",
 666 |        "    <tr>\n",
 667 |        "      <th>7</th>\n",
 668 |        "      <td>37.73</td>\n",
 669 |        "      <td>1</td>\n",
 670 |        "      <td>100.0</td>\n",
 671 |        "      <td>30.375</td>\n",
 672 |        "      <td>2.32815</td>\n",
 673 |        "      <td>7.63431</td>\n",
 674 |        "      <td>5.70686</td>\n",
 675 |        "      <td>2.31663</td>\n",
 676 |        "      <td>7.64745</td>\n",
 677 |        "      <td>6.01495</td>\n",
 678 |        "      <td>...</td>\n",
 679 |        "      <td>9.61685</td>\n",
 680 |        "      <td>-1.76749</td>\n",
 681 |        "      <td>0.326380</td>\n",
 682 |        "      <td>-0.010352</td>\n",
 683 |        "      <td>-0.016621</td>\n",
 684 |        "      <td>0.006548</td>\n",
 685 |        "      <td>-61.5738</td>\n",
 686 |        "      <td>-36.1724</td>\n",
 687 |        "      <td>-59.3487</td>\n",
 688 |        "      <td>1</td>\n",
 689 |        "    </tr>\n",
 690 |        "    <tr>\n",
 691 |        "      <th>8</th>\n",
 692 |        "      <td>37.74</td>\n",
 693 |        "      <td>1</td>\n",
 694 |        "      <td>100.0</td>\n",
 695 |        "      <td>30.375</td>\n",
 696 |        "      <td>2.25096</td>\n",
 697 |        "      <td>7.78598</td>\n",
 698 |        "      <td>5.62821</td>\n",
 699 |        "      <td>2.28637</td>\n",
 700 |        "      <td>7.70801</td>\n",
 701 |        "      <td>5.93935</td>\n",
 702 |        "      <td>...</td>\n",
 703 |        "      <td>9.61686</td>\n",
 704 |        "      <td>-1.72212</td>\n",
 705 |        "      <td>0.326234</td>\n",
 706 |        "      <td>0.039346</td>\n",
 707 |        "      <td>0.020393</td>\n",
 708 |        "      <td>-0.011880</td>\n",
 709 |        "      <td>-61.7741</td>\n",
 710 |        "      <td>-37.1744</td>\n",
 711 |        "      <td>-58.1199</td>\n",
 712 |        "      <td>1</td>\n",
 713 |        "    </tr>\n",
 714 |        "    <tr>\n",
 715 |        "      <th>9</th>\n",
 716 |        "      <td>37.75</td>\n",
 717 |        "      <td>1</td>\n",
 718 |        "      <td>100.0</td>\n",
 719 |        "      <td>30.375</td>\n",
 720 |        "      <td>2.14107</td>\n",
 721 |        "      <td>7.52262</td>\n",
 722 |        "      <td>5.78141</td>\n",
 723 |        "      <td>2.31538</td>\n",
 724 |        "      <td>7.72276</td>\n",
 725 |        "      <td>5.78828</td>\n",
 726 |        "      <td>...</td>\n",
 727 |        "      <td>9.63189</td>\n",
 728 |        "      <td>-1.70699</td>\n",
 729 |        "      <td>0.326105</td>\n",
 730 |        "      <td>0.029874</td>\n",
 731 |        "      <td>-0.010763</td>\n",
 732 |        "      <td>0.005133</td>\n",
 733 |        "      <td>-60.7680</td>\n",
 734 |        "      <td>-37.4206</td>\n",
 735 |        "      <td>-58.8735</td>\n",
 736 |        "      <td>1</td>\n",
 737 |        "    </tr>\n",
 738 |        "  </tbody>\n",
 739 |        "</table>\n",
 740 |        "<p>10 rows × 43 columns</p>\n",
 741 |        "</div>"
 742 |       ],
 743 |       "text/plain": [
 744 |        "   timestamp  activityID  heartrate  handTemperature  handAcc16_1  \\\n",
 745 |        "0      37.66           1        NaN           30.375      2.21530   \n",
 746 |        "1      37.67           1        NaN           30.375      2.29196   \n",
 747 |        "2      37.68           1        NaN           30.375      2.29090   \n",
 748 |        "3      37.69           1        NaN           30.375      2.21800   \n",
 749 |        "4      37.70           1      100.0           30.375      2.30106   \n",
 750 |        "5      37.71           1      100.0           30.375      2.07165   \n",
 751 |        "6      37.72           1      100.0           30.375      2.41148   \n",
 752 |        "7      37.73           1      100.0           30.375      2.32815   \n",
 753 |        "8      37.74           1      100.0           30.375      2.25096   \n",
 754 |        "9      37.75           1      100.0           30.375      2.14107   \n",
 755 |        "\n",
 756 |        "   handAcc16_2  handAcc16_3  handAcc6_1  handAcc6_2  handAcc6_3     ...      \\\n",
 757 |        "0      8.27915      5.58753     2.24689     8.55387     5.77143     ...       \n",
 758 |        "1      7.67288      5.74467     2.27373     8.14592     5.78739     ...       \n",
 759 |        "2      7.14240      5.82342     2.26966     7.66268     5.78846     ...       \n",
 760 |        "3      7.14365      5.89930     2.22177     7.25535     5.88000     ...       \n",
 761 |        "4      7.25857      6.09259     2.20720     7.24042     5.95555     ...       \n",
 762 |        "5      7.25965      6.01218     2.19238     7.21038     6.01604     ...       \n",
 763 |        "6      7.59780      5.93915     2.23988     7.46679     6.03053     ...       \n",
 764 |        "7      7.63431      5.70686     2.31663     7.64745     6.01495     ...       \n",
 765 |        "8      7.78598      5.62821     2.28637     7.70801     5.93935     ...       \n",
 766 |        "9      7.52262      5.78141     2.31538     7.72276     5.78828     ...       \n",
 767 |        "\n",
 768 |        "   ankleAcc6_1  ankleAcc6_2  ankleAcc6_3  ankleGyro1  ankleGyro2  ankleGyro3  \\\n",
 769 |        "0      9.63162     -1.76757     0.265761    0.002908   -0.027714    0.001752   \n",
 770 |        "1      9.58649     -1.75247     0.250816    0.020882    0.000945    0.006007   \n",
 771 |        "2      9.60196     -1.73721     0.356632   -0.035392   -0.052422   -0.004882   \n",
 772 |        "3      9.58674     -1.78264     0.311453   -0.032514   -0.018844    0.026950   \n",
 773 |        "4      9.64677     -1.75240     0.295902    0.001351   -0.048878   -0.006328   \n",
 774 |        "5      9.60177     -1.75239     0.311276    0.003793   -0.026906    0.004125   \n",
 775 |        "6      9.67694     -1.76748     0.326060    0.036814   -0.032277   -0.006866   \n",
 776 |        "7      9.61685     -1.76749     0.326380   -0.010352   -0.016621    0.006548   \n",
 777 |        "8      9.61686     -1.72212     0.326234    0.039346    0.020393   -0.011880   \n",
 778 |        "9      9.63189     -1.70699     0.326105    0.029874   -0.010763    0.005133   \n",
 779 |        "\n",
 780 |        "   ankleMagne1  ankleMagne2  ankleMagne3  subject_id  \n",
 781 |        "0     -61.1081     -36.8636     -58.3696           1  \n",
 782 |        "1     -60.8916     -36.3197     -58.3656           1  \n",
 783 |        "2     -60.3407     -35.7842     -58.6119           1  \n",
 784 |        "3     -60.7646     -37.1028     -57.8799           1  \n",
 785 |        "4     -60.2040     -37.1225     -57.8847           1  \n",
 786 |        "5     -61.3257     -36.9744     -57.7501           1  \n",
 787 |        "6     -61.5520     -36.9632     -57.9957           1  \n",
 788 |        "7     -61.5738     -36.1724     -59.3487           1  \n",
 789 |        "8     -61.7741     -37.1744     -58.1199           1  \n",
 790 |        "9     -60.7680     -37.4206     -58.8735           1  \n",
 791 |        "\n",
 792 |        "[10 rows x 43 columns]"
 793 |       ]
 794 |      },
 795 |      "execution_count": 6,
 796 |      "metadata": {},
 797 |      "output_type": "execute_result"
 798 |     }
 799 |    ],
 800 |    "source": [
 801 |     "dataCol.reset_index(drop = True, inplace = True)\n",
 802 |     "dataCol.head(10)"
 803 |    ]
 804 |   },
 805 |   {
 806 |    "cell_type": "markdown",
 807 |    "metadata": {},
 808 |    "source": [
 809 |     "The reason why heartrate still has NaN values is because interpolation calculates the values around the NaN cell. Since the first cells are NaN it is normal to generate new NaN values after interpolation. To overcome this problem we can assume that the value of the first 4 cells is 100 since the values after the index 4 is 100. Doing so will eliminate any NaN values from our dataset."
 810 |    ]
 811 |   },
 812 |   {
 813 |    "cell_type": "code",
 814 |    "execution_count": 7,
 815 |    "metadata": {},
 816 |    "outputs": [
 817 |     {
 818 |      "data": {
 819 |       "text/plain": [
 820 |        "timestamp           0\n",
 821 |        "activityID          0\n",
 822 |        "heartrate           4\n",
 823 |        "handTemperature     0\n",
 824 |        "handAcc16_1         0\n",
 825 |        "handAcc16_2         0\n",
 826 |        "handAcc16_3         0\n",
 827 |        "handAcc6_1          0\n",
 828 |        "handAcc6_2          0\n",
 829 |        "handAcc6_3          0\n",
 830 |        "handGyro1           0\n",
 831 |        "handGyro2           0\n",
 832 |        "handGyro3           0\n",
 833 |        "handMagne1          0\n",
 834 |        "handMagne2          0\n",
 835 |        "handMagne3          0\n",
 836 |        "chestTemperature    0\n",
 837 |        "chestAcc16_1        0\n",
 838 |        "chestAcc16_2        0\n",
 839 |        "chestAcc16_3        0\n",
 840 |        "chestAcc6_1         0\n",
 841 |        "chestAcc6_2         0\n",
 842 |        "chestAcc6_3         0\n",
 843 |        "chestGyro1          0\n",
 844 |        "chestGyro2          0\n",
 845 |        "chestGyro3          0\n",
 846 |        "chestMagne1         0\n",
 847 |        "chestMagne2         0\n",
 848 |        "chestMagne3         0\n",
 849 |        "ankleTemperature    0\n",
 850 |        "ankleAcc16_1        0\n",
 851 |        "ankleAcc16_2        0\n",
 852 |        "ankleAcc16_3        0\n",
 853 |        "ankleAcc6_1         0\n",
 854 |        "ankleAcc6_2         0\n",
 855 |        "ankleAcc6_3         0\n",
 856 |        "ankleGyro1          0\n",
 857 |        "ankleGyro2          0\n",
 858 |        "ankleGyro3          0\n",
 859 |        "ankleMagne1         0\n",
 860 |        "ankleMagne2         0\n",
 861 |        "ankleMagne3         0\n",
 862 |        "subject_id          0\n",
 863 |        "dtype: int64"
 864 |       ]
 865 |      },
 866 |      "execution_count": 7,
 867 |      "metadata": {},
 868 |      "output_type": "execute_result"
 869 |     }
 870 |    ],
 871 |    "source": [
 872 |     "dataCol.isnull().sum()"
 873 |    ]
 874 |   },
 875 |   {
 876 |    "cell_type": "markdown",
 877 |    "metadata": {},
 878 |    "source": [
 879 |     "The reason why HR still has nan values is because the interpolation calculates the values around the nan cell. Since the first cells are nan it is normal to generate new nan values after interpolation. To overcome this problem we can assume that the value of the first 4 cells is 100 since the values after the index 4 is 100."
 880 |    ]
 881 |   },
 882 |   {
 883 |    "cell_type": "code",
 884 |    "execution_count": 8,
 885 |    "metadata": {},
 886 |    "outputs": [
 887 |     {
 888 |      "name": "stderr",
 889 |      "output_type": "stream",
 890 |      "text": [
 891 |       "C:\\Users\\Andreas\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\pandas\\core\\indexing.py:189: SettingWithCopyWarning: \n",
 892 |       "A value is trying to be set on a copy of a slice from a DataFrame\n",
 893 |       "\n",
 894 |       "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
 895 |       "  self._setitem_with_indexer(indexer, value)\n"
 896 |      ]
 897 |     }
 898 |    ],
 899 |    "source": [
 900 |     "for i in range(0,4):\n",
 901 |     "    dataCol[\"heartrate\"].iloc[i]=100"
 902 |    ]
 903 |   },
 904 |   {
 905 |    "cell_type": "markdown",
 906 |    "metadata": {},
 907 |    "source": [
 908 |     "It is clear that there are no missing values but we can check just to be sure. "
 909 |    ]
 910 |   },
 911 |   {
 912 |    "cell_type": "code",
 913 |    "execution_count": 9,
 914 |    "metadata": {},
 915 |    "outputs": [
 916 |     {
 917 |      "data": {
 918 |       "text/plain": [
 919 |        "timestamp           0\n",
 920 |        "activityID          0\n",
 921 |        "heartrate           0\n",
 922 |        "handTemperature     0\n",
 923 |        "handAcc16_1         0\n",
 924 |        "handAcc16_2         0\n",
 925 |        "handAcc16_3         0\n",
 926 |        "handAcc6_1          0\n",
 927 |        "handAcc6_2          0\n",
 928 |        "handAcc6_3          0\n",
 929 |        "handGyro1           0\n",
 930 |        "handGyro2           0\n",
 931 |        "handGyro3           0\n",
 932 |        "handMagne1          0\n",
 933 |        "handMagne2          0\n",
 934 |        "handMagne3          0\n",
 935 |        "chestTemperature    0\n",
 936 |        "chestAcc16_1        0\n",
 937 |        "chestAcc16_2        0\n",
 938 |        "chestAcc16_3        0\n",
 939 |        "chestAcc6_1         0\n",
 940 |        "chestAcc6_2         0\n",
 941 |        "chestAcc6_3         0\n",
 942 |        "chestGyro1          0\n",
 943 |        "chestGyro2          0\n",
 944 |        "chestGyro3          0\n",
 945 |        "chestMagne1         0\n",
 946 |        "chestMagne2         0\n",
 947 |        "chestMagne3         0\n",
 948 |        "ankleTemperature    0\n",
 949 |        "ankleAcc16_1        0\n",
 950 |        "ankleAcc16_2        0\n",
 951 |        "ankleAcc16_3        0\n",
 952 |        "ankleAcc6_1         0\n",
 953 |        "ankleAcc6_2         0\n",
 954 |        "ankleAcc6_3         0\n",
 955 |        "ankleGyro1          0\n",
 956 |        "ankleGyro2          0\n",
 957 |        "ankleGyro3          0\n",
 958 |        "ankleMagne1         0\n",
 959 |        "ankleMagne2         0\n",
 960 |        "ankleMagne3         0\n",
 961 |        "subject_id          0\n",
 962 |        "dtype: int64"
 963 |       ]
 964 |      },
 965 |      "execution_count": 9,
 966 |      "metadata": {},
 967 |      "output_type": "execute_result"
 968 |     }
 969 |    ],
 970 |    "source": [
 971 |     "dataCol.isnull().sum()"
 972 |    ]
 973 |   },
 974 |   {
 975 |    "cell_type": "markdown",
 976 |    "metadata": {},
 977 |    "source": [
 978 |     "As we can see, there are no more missing values. We can now move into Exploratory Data Analysis."
 979 |    ]
 980 |   },
 981 |   {
 982 |    "cell_type": "markdown",
 983 |    "metadata": {},
 984 |    "source": [
 985 |     "## Exploratory Data Analysis"
 986 |    ]
 987 |   },
 988 |   {
 989 |    "cell_type": "markdown",
 990 |    "metadata": {},
 991 |    "source": [
 992 |     "#### Splitting Data in Train and Test sets"
 993 |    ]
 994 |   },
 995 |   {
 996 |    "cell_type": "markdown",
 997 |    "metadata": {},
 998 |    "source": [
 999 |     "Before splitting our data, it is best if we first check if the classes are balanced which would imply that stratification is not needed for our split. If classes' weights are unbalanced, we should proceed with stratifying while splitting the data. Stratified sampling is the process of taking samples of all classes and putting them on the sub-sets created for train and test sets. It helps with getting better results when the distribution of the weights of the classes is not normal. Normal distribution would imply that each class' weight is similar with all other weights of classes. "
1000 |    ]
1001 |   },
1002 |   {
1003 |    "cell_type": "markdown",
1004 |    "metadata": {},
1005 |    "source": [
1006 |     "## See class distribution/ration\n",
1007 |     "Check if our dataset is balanced"
1008 |    ]
1009 |   },
1010 |   {
1011 |    "cell_type": "code",
1012 |    "execution_count": 10,
1013 |    "metadata": {},
1014 |    "outputs": [
1015 |     {
1016 |      "data": {
1017 |       "image/png": "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\n",
1018 |       "text/plain": [
1019 |        "<Figure size 864x432 with 1 Axes>"
1020 |       ]
1021 |      },
1022 |      "metadata": {},
1023 |      "output_type": "display_data"
1024 |     }
1025 |    ],
1026 |    "source": [
1027 |     "dataCol['activityID'].value_counts().plot(kind = \"bar\",figsize = (12,6))\n",
1028 |     "plt.show()"
1029 |    ]
1030 |   },
1031 |   {
1032 |    "cell_type": "markdown",
1033 |    "metadata": {},
1034 |    "source": [
1035 |     "As the above plot shows, our classes are mostly balanced. Therefore, we proceed with splitting the data into train and test sets. The most common split in the industry is a 80% split for train set and 20% for test set, which is the split fraction that will be used. "
1036 |    ]
1037 |   },
1038 |   {
1039 |    "cell_type": "code",
1040 |    "execution_count": 11,
1041 |    "metadata": {},
1042 |    "outputs": [],
1043 |    "source": [
1044 |     "train_df = dataCol.sample(frac=0.8, random_state=1)\n",
1045 |     "test_df = dataCol.drop(train_df.index)"
1046 |    ]
1047 |   },
1048 |   {
1049 |    "cell_type": "markdown",
1050 |    "metadata": {},
1051 |    "source": [
1052 |     "We need to check some statistics/insights from the describe method from pandas as it can provide further results that we need to know as to how our data is being processed  for our analysis."
1053 |    ]
1054 |   },
1055 |   {
1056 |    "cell_type": "code",
1057 |    "execution_count": 12,
1058 |    "metadata": {},
1059 |    "outputs": [
1060 |     {
1061 |      "data": {
1062 |       "text/html": [
1063 |        "<div>\n",
1064 |        "<style scoped>\n",
1065 |        "    .dataframe tbody tr th:only-of-type {\n",
1066 |        "        vertical-align: middle;\n",
1067 |        "    }\n",
1068 |        "\n",
1069 |        "    .dataframe tbody tr th {\n",
1070 |        "        vertical-align: top;\n",
1071 |        "    }\n",
1072 |        "\n",
1073 |        "    .dataframe thead th {\n",
1074 |        "        text-align: right;\n",
1075 |        "    }\n",
1076 |        "</style>\n",
1077 |        "<table border=\"1\" class=\"dataframe\">\n",
1078 |        "  <thead>\n",
1079 |        "    <tr style=\"text-align: right;\">\n",
1080 |        "      <th></th>\n",
1081 |        "      <th>timestamp</th>\n",
1082 |        "      <th>activityID</th>\n",
1083 |        "      <th>heartrate</th>\n",
1084 |        "      <th>handTemperature</th>\n",
1085 |        "      <th>handAcc16_1</th>\n",
1086 |        "      <th>handAcc16_2</th>\n",
1087 |        "      <th>handAcc16_3</th>\n",
1088 |        "      <th>handAcc6_1</th>\n",
1089 |        "      <th>handAcc6_2</th>\n",
1090 |        "      <th>handAcc6_3</th>\n",
1091 |        "      <th>...</th>\n",
1092 |        "      <th>ankleAcc6_1</th>\n",
1093 |        "      <th>ankleAcc6_2</th>\n",
1094 |        "      <th>ankleAcc6_3</th>\n",
1095 |        "      <th>ankleGyro1</th>\n",
1096 |        "      <th>ankleGyro2</th>\n",
1097 |        "      <th>ankleGyro3</th>\n",
1098 |        "      <th>ankleMagne1</th>\n",
1099 |        "      <th>ankleMagne2</th>\n",
1100 |        "      <th>ankleMagne3</th>\n",
1101 |        "      <th>subject_id</th>\n",
1102 |        "    </tr>\n",
1103 |        "  </thead>\n",
1104 |        "  <tbody>\n",
1105 |        "    <tr>\n",
1106 |        "      <th>count</th>\n",
1107 |        "      <td>1.554298e+06</td>\n",
1108 |        "      <td>1.554298e+06</td>\n",
1109 |        "      <td>1.554298e+06</td>\n",
1110 |        "      <td>1.554298e+06</td>\n",
1111 |        "      <td>1.554298e+06</td>\n",
1112 |        "      <td>1.554298e+06</td>\n",
1113 |        "      <td>1.554298e+06</td>\n",
1114 |        "      <td>1.554298e+06</td>\n",
1115 |        "      <td>1.554298e+06</td>\n",
1116 |        "      <td>1.554298e+06</td>\n",
1117 |        "      <td>...</td>\n",
1118 |        "      <td>1.554298e+06</td>\n",
1119 |        "      <td>1.554298e+06</td>\n",
1120 |        "      <td>1.554298e+06</td>\n",
1121 |        "      <td>1.554298e+06</td>\n",
1122 |        "      <td>1.554298e+06</td>\n",
1123 |        "      <td>1.554298e+06</td>\n",
1124 |        "      <td>1.554298e+06</td>\n",
1125 |        "      <td>1.554298e+06</td>\n",
1126 |        "      <td>1.554298e+06</td>\n",
1127 |        "      <td>1.554298e+06</td>\n",
1128 |        "    </tr>\n",
1129 |        "    <tr>\n",
1130 |        "      <th>mean</th>\n",
1131 |        "      <td>1.705049e+03</td>\n",
1132 |        "      <td>8.080534e+00</td>\n",
1133 |        "      <td>1.074758e+02</td>\n",
1134 |        "      <td>3.275138e+01</td>\n",
1135 |        "      <td>-4.953526e+00</td>\n",
1136 |        "      <td>3.581113e+00</td>\n",
1137 |        "      <td>3.603514e+00</td>\n",
1138 |        "      <td>-4.886385e+00</td>\n",
1139 |        "      <td>3.570863e+00</td>\n",
1140 |        "      <td>3.787763e+00</td>\n",
1141 |        "      <td>...</td>\n",
1142 |        "      <td>9.374706e+00</td>\n",
1143 |        "      <td>-4.445267e-02</td>\n",
1144 |        "      <td>-2.175620e+00</td>\n",
1145 |        "      <td>1.027736e-02</td>\n",
1146 |        "      <td>-3.649406e-02</td>\n",
1147 |        "      <td>5.607242e-03</td>\n",
1148 |        "      <td>-3.157962e+01</td>\n",
1149 |        "      <td>1.394841e+00</td>\n",
1150 |        "      <td>1.725137e+01</td>\n",
1151 |        "      <td>4.566578e+00</td>\n",
1152 |        "    </tr>\n",
1153 |        "    <tr>\n",
1154 |        "      <th>std</th>\n",
1155 |        "      <td>1.093592e+03</td>\n",
1156 |        "      <td>6.175064e+00</td>\n",
1157 |        "      <td>2.699031e+01</td>\n",
1158 |        "      <td>1.794207e+00</td>\n",
1159 |        "      <td>6.239143e+00</td>\n",
1160 |        "      <td>6.886169e+00</td>\n",
1161 |        "      <td>3.958145e+00</td>\n",
1162 |        "      <td>6.245060e+00</td>\n",
1163 |        "      <td>6.585066e+00</td>\n",
1164 |        "      <td>3.945398e+00</td>\n",
1165 |        "      <td>...</td>\n",
1166 |        "      <td>6.067489e+00</td>\n",
1167 |        "      <td>7.183548e+00</td>\n",
1168 |        "      <td>3.475628e+00</td>\n",
1169 |        "      <td>1.126197e+00</td>\n",
1170 |        "      <td>6.380780e-01</td>\n",
1171 |        "      <td>2.011908e+00</td>\n",
1172 |        "      <td>1.834688e+01</td>\n",
1173 |        "      <td>2.168353e+01</td>\n",
1174 |        "      <td>1.969368e+01</td>\n",
1175 |        "      <td>2.333375e+00</td>\n",
1176 |        "    </tr>\n",
1177 |        "    <tr>\n",
1178 |        "      <th>min</th>\n",
1179 |        "      <td>3.120000e+01</td>\n",
1180 |        "      <td>1.000000e+00</td>\n",
1181 |        "      <td>5.700000e+01</td>\n",
1182 |        "      <td>2.487500e+01</td>\n",
1183 |        "      <td>-1.453670e+02</td>\n",
1184 |        "      <td>-1.043010e+02</td>\n",
1185 |        "      <td>-1.014520e+02</td>\n",
1186 |        "      <td>-6.121470e+01</td>\n",
1187 |        "      <td>-6.184170e+01</td>\n",
1188 |        "      <td>-6.193470e+01</td>\n",
1189 |        "      <td>...</td>\n",
1190 |        "      <td>-6.114200e+01</td>\n",
1191 |        "      <td>-6.190350e+01</td>\n",
1192 |        "      <td>-6.231480e+01</td>\n",
1193 |        "      <td>-1.416200e+01</td>\n",
1194 |        "      <td>-1.304010e+01</td>\n",
1195 |        "      <td>-1.401960e+01</td>\n",
1196 |        "      <td>-1.726240e+02</td>\n",
1197 |        "      <td>-1.379080e+02</td>\n",
1198 |        "      <td>-1.027160e+02</td>\n",
1199 |        "      <td>1.000000e+00</td>\n",
1200 |        "    </tr>\n",
1201 |        "    <tr>\n",
1202 |        "      <th>25%</th>\n",
1203 |        "      <td>7.442925e+02</td>\n",
1204 |        "      <td>3.000000e+00</td>\n",
1205 |        "      <td>8.600000e+01</td>\n",
1206 |        "      <td>3.168750e+01</td>\n",
1207 |        "      <td>-8.970020e+00</td>\n",
1208 |        "      <td>1.057830e+00</td>\n",
1209 |        "      <td>1.162090e+00</td>\n",
1210 |        "      <td>-8.867070e+00</td>\n",
1211 |        "      <td>1.055563e+00</td>\n",
1212 |        "      <td>1.365070e+00</td>\n",
1213 |        "      <td>...</td>\n",
1214 |        "      <td>8.396590e+00</td>\n",
1215 |        "      <td>-2.073120e+00</td>\n",
1216 |        "      <td>-3.399390e+00</td>\n",
1217 |        "      <td>-2.081648e-01</td>\n",
1218 |        "      <td>-1.066463e-01</td>\n",
1219 |        "      <td>-4.416657e-01</td>\n",
1220 |        "      <td>-4.170160e+01</td>\n",
1221 |        "      <td>-1.246927e+01</td>\n",
1222 |        "      <td>3.799633e+00</td>\n",
1223 |        "      <td>2.000000e+00</td>\n",
1224 |        "    </tr>\n",
1225 |        "    <tr>\n",
1226 |        "      <th>50%</th>\n",
1227 |        "      <td>1.480090e+03</td>\n",
1228 |        "      <td>6.000000e+00</td>\n",
1229 |        "      <td>1.040000e+02</td>\n",
1230 |        "      <td>3.312500e+01</td>\n",
1231 |        "      <td>-5.449130e+00</td>\n",
1232 |        "      <td>3.525300e+00</td>\n",
1233 |        "      <td>3.432840e+00</td>\n",
1234 |        "      <td>-5.377104e+00</td>\n",
1235 |        "      <td>3.566820e+00</td>\n",
1236 |        "      <td>3.663470e+00</td>\n",
1237 |        "      <td>...</td>\n",
1238 |        "      <td>9.550020e+00</td>\n",
1239 |        "      <td>-2.252810e-01</td>\n",
1240 |        "      <td>-1.993145e+00</td>\n",
1241 |        "      <td>4.636280e-03</td>\n",
1242 |        "      <td>-3.977450e-03</td>\n",
1243 |        "      <td>-2.336400e-03</td>\n",
1244 |        "      <td>-3.400060e+01</td>\n",
1245 |        "      <td>7.672570e-01</td>\n",
1246 |        "      <td>1.876795e+01</td>\n",
1247 |        "      <td>5.000000e+00</td>\n",
1248 |        "    </tr>\n",
1249 |        "    <tr>\n",
1250 |        "      <th>75%</th>\n",
1251 |        "      <td>2.664000e+03</td>\n",
1252 |        "      <td>1.300000e+01</td>\n",
1253 |        "      <td>1.240000e+02</td>\n",
1254 |        "      <td>3.406250e+01</td>\n",
1255 |        "      <td>-9.581008e-01</td>\n",
1256 |        "      <td>6.450507e+00</td>\n",
1257 |        "      <td>6.532445e+00</td>\n",
1258 |        "      <td>-9.061720e-01</td>\n",
1259 |        "      <td>6.458267e+00</td>\n",
1260 |        "      <td>6.778200e+00</td>\n",
1261 |        "      <td>...</td>\n",
1262 |        "      <td>1.028160e+01</td>\n",
1263 |        "      <td>1.920960e+00</td>\n",
1264 |        "      <td>-5.958823e-01</td>\n",
1265 |        "      <td>1.308070e-01</td>\n",
1266 |        "      <td>1.160640e-01</td>\n",
1267 |        "      <td>9.121635e-02</td>\n",
1268 |        "      <td>-1.789610e+01</td>\n",
1269 |        "      <td>1.782927e+01</td>\n",
1270 |        "      <td>3.120910e+01</td>\n",
1271 |        "      <td>7.000000e+00</td>\n",
1272 |        "    </tr>\n",
1273 |        "    <tr>\n",
1274 |        "      <th>max</th>\n",
1275 |        "      <td>4.245680e+03</td>\n",
1276 |        "      <td>2.400000e+01</td>\n",
1277 |        "      <td>2.020000e+02</td>\n",
1278 |        "      <td>3.550000e+01</td>\n",
1279 |        "      <td>6.285960e+01</td>\n",
1280 |        "      <td>1.556990e+02</td>\n",
1281 |        "      <td>1.577600e+02</td>\n",
1282 |        "      <td>5.282140e+01</td>\n",
1283 |        "      <td>6.225980e+01</td>\n",
1284 |        "      <td>6.192340e+01</td>\n",
1285 |        "      <td>...</td>\n",
1286 |        "      <td>6.196930e+01</td>\n",
1287 |        "      <td>6.204900e+01</td>\n",
1288 |        "      <td>6.093570e+01</td>\n",
1289 |        "      <td>1.742040e+01</td>\n",
1290 |        "      <td>1.358820e+01</td>\n",
1291 |        "      <td>1.448270e+01</td>\n",
1292 |        "      <td>9.155160e+01</td>\n",
1293 |        "      <td>9.369920e+01</td>\n",
1294 |        "      <td>1.469000e+02</td>\n",
1295 |        "      <td>9.000000e+00</td>\n",
1296 |        "    </tr>\n",
1297 |        "  </tbody>\n",
1298 |        "</table>\n",
1299 |        "<p>8 rows × 43 columns</p>\n",
1300 |        "</div>"
1301 |       ],
1302 |       "text/plain": [
1303 |        "          timestamp    activityID     heartrate  handTemperature  \\\n",
1304 |        "count  1.554298e+06  1.554298e+06  1.554298e+06     1.554298e+06   \n",
1305 |        "mean   1.705049e+03  8.080534e+00  1.074758e+02     3.275138e+01   \n",
1306 |        "std    1.093592e+03  6.175064e+00  2.699031e+01     1.794207e+00   \n",
1307 |        "min    3.120000e+01  1.000000e+00  5.700000e+01     2.487500e+01   \n",
1308 |        "25%    7.442925e+02  3.000000e+00  8.600000e+01     3.168750e+01   \n",
1309 |        "50%    1.480090e+03  6.000000e+00  1.040000e+02     3.312500e+01   \n",
1310 |        "75%    2.664000e+03  1.300000e+01  1.240000e+02     3.406250e+01   \n",
1311 |        "max    4.245680e+03  2.400000e+01  2.020000e+02     3.550000e+01   \n",
1312 |        "\n",
1313 |        "        handAcc16_1   handAcc16_2   handAcc16_3    handAcc6_1    handAcc6_2  \\\n",
1314 |        "count  1.554298e+06  1.554298e+06  1.554298e+06  1.554298e+06  1.554298e+06   \n",
1315 |        "mean  -4.953526e+00  3.581113e+00  3.603514e+00 -4.886385e+00  3.570863e+00   \n",
1316 |        "std    6.239143e+00  6.886169e+00  3.958145e+00  6.245060e+00  6.585066e+00   \n",
1317 |        "min   -1.453670e+02 -1.043010e+02 -1.014520e+02 -6.121470e+01 -6.184170e+01   \n",
1318 |        "25%   -8.970020e+00  1.057830e+00  1.162090e+00 -8.867070e+00  1.055563e+00   \n",
1319 |        "50%   -5.449130e+00  3.525300e+00  3.432840e+00 -5.377104e+00  3.566820e+00   \n",
1320 |        "75%   -9.581008e-01  6.450507e+00  6.532445e+00 -9.061720e-01  6.458267e+00   \n",
1321 |        "max    6.285960e+01  1.556990e+02  1.577600e+02  5.282140e+01  6.225980e+01   \n",
1322 |        "\n",
1323 |        "         handAcc6_3      ...        ankleAcc6_1   ankleAcc6_2   ankleAcc6_3  \\\n",
1324 |        "count  1.554298e+06      ...       1.554298e+06  1.554298e+06  1.554298e+06   \n",
1325 |        "mean   3.787763e+00      ...       9.374706e+00 -4.445267e-02 -2.175620e+00   \n",
1326 |        "std    3.945398e+00      ...       6.067489e+00  7.183548e+00  3.475628e+00   \n",
1327 |        "min   -6.193470e+01      ...      -6.114200e+01 -6.190350e+01 -6.231480e+01   \n",
1328 |        "25%    1.365070e+00      ...       8.396590e+00 -2.073120e+00 -3.399390e+00   \n",
1329 |        "50%    3.663470e+00      ...       9.550020e+00 -2.252810e-01 -1.993145e+00   \n",
1330 |        "75%    6.778200e+00      ...       1.028160e+01  1.920960e+00 -5.958823e-01   \n",
1331 |        "max    6.192340e+01      ...       6.196930e+01  6.204900e+01  6.093570e+01   \n",
1332 |        "\n",
1333 |        "         ankleGyro1    ankleGyro2    ankleGyro3   ankleMagne1   ankleMagne2  \\\n",
1334 |        "count  1.554298e+06  1.554298e+06  1.554298e+06  1.554298e+06  1.554298e+06   \n",
1335 |        "mean   1.027736e-02 -3.649406e-02  5.607242e-03 -3.157962e+01  1.394841e+00   \n",
1336 |        "std    1.126197e+00  6.380780e-01  2.011908e+00  1.834688e+01  2.168353e+01   \n",
1337 |        "min   -1.416200e+01 -1.304010e+01 -1.401960e+01 -1.726240e+02 -1.379080e+02   \n",
1338 |        "25%   -2.081648e-01 -1.066463e-01 -4.416657e-01 -4.170160e+01 -1.246927e+01   \n",
1339 |        "50%    4.636280e-03 -3.977450e-03 -2.336400e-03 -3.400060e+01  7.672570e-01   \n",
1340 |        "75%    1.308070e-01  1.160640e-01  9.121635e-02 -1.789610e+01  1.782927e+01   \n",
1341 |        "max    1.742040e+01  1.358820e+01  1.448270e+01  9.155160e+01  9.369920e+01   \n",
1342 |        "\n",
1343 |        "        ankleMagne3    subject_id  \n",
1344 |        "count  1.554298e+06  1.554298e+06  \n",
1345 |        "mean   1.725137e+01  4.566578e+00  \n",
1346 |        "std    1.969368e+01  2.333375e+00  \n",
1347 |        "min   -1.027160e+02  1.000000e+00  \n",
1348 |        "25%    3.799633e+00  2.000000e+00  \n",
1349 |        "50%    1.876795e+01  5.000000e+00  \n",
1350 |        "75%    3.120910e+01  7.000000e+00  \n",
1351 |        "max    1.469000e+02  9.000000e+00  \n",
1352 |        "\n",
1353 |        "[8 rows x 43 columns]"
1354 |       ]
1355 |      },
1356 |      "execution_count": 12,
1357 |      "metadata": {},
1358 |      "output_type": "execute_result"
1359 |     }
1360 |    ],
1361 |    "source": [
1362 |     "train_df.describe()"
1363 |    ]
1364 |   },
1365 |   {
1366 |    "cell_type": "markdown",
1367 |    "metadata": {},
1368 |    "source": [
1369 |     "We are going to focus on heart rate as it is our most precice meter of check for tracking subjects during activities as implied by the various indications on the readme file of the dataset. With that in mind, looking at the table, we can observe that the mean heart rate throughout the dataset is 107.4 . Furthermore the minimum heart rate is 57 and the maximum heart rate is 202. The quartiles that are shown can be further analysed by plotting a box plot which will help with understanding our outliers and quartiles groups and also shown the mean of our data's heart rate."
1370 |    ]
1371 |   },
1372 |   {
1373 |    "cell_type": "code",
1374 |    "execution_count": 13,
1375 |    "metadata": {},
1376 |    "outputs": [
1377 |     {
1378 |      "data": {
1379 |       "image/png": "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\n",
1380 |       "text/plain": [
1381 |        "<Figure size 288x288 with 1 Axes>"
1382 |       ]
1383 |      },
1384 |      "metadata": {},
1385 |      "output_type": "display_data"
1386 |     }
1387 |    ],
1388 |    "source": [
1389 |     "import seaborn as sns\n",
1390 |     "\n",
1391 |     "fig, ax = plt.subplots(figsize=(4,4))\n",
1392 |     "plt.title(\"Heart Rate \")\n",
1393 |     "ax = sns.boxplot(y=train_df[\"heartrate\"])"
1394 |    ]
1395 |   },
1396 |   {
1397 |    "cell_type": "markdown",
1398 |    "metadata": {},
1399 |    "source": [
1400 |     "Looking at the box plot, we can see that the outliers have heart rate from 180 up to 202. Our highest quartile group out of the four starts from 124 which is the end of the Inter-quartile range and finishes at 180 which also makes it our biggest group by looking at the size of it on the box plot compared to the other quartiles. Meaning that the biggest amount of subjects on the activities performed had heart rate of 124 up to 180. Our third quartile group starts from the mean value which is 107.4 which is where the horizontal line in our box is, and finishes at the end of the Inter-quartile range which is 124. Our second quartile group, starts from the start of the Inter-quartile range which is 86 and ends at the mean value 124. Our first quartile group starts from the lowest data point, 57 and ends at the start of the Inter-quartile range 86. Our box plot also shows that most subjects performed some activities at a statistically similar way but failed to do the same in all activities which explains the big upper quartile group.\n",
1401 |     "\n",
1402 |     "To find the most cumbersome activities we have to plot a bar chart which will show the mean values of heart rate for each activity that was performed. This will in return enable us to analyse further specific activity data. The names of the activities will be used for easier analysis of results."
1403 |    ]
1404 |   },
1405 |   {
1406 |    "cell_type": "code",
1407 |    "execution_count": 15,
1408 |    "metadata": {},
1409 |    "outputs": [
1410 |     {
1411 |      "data": {
1412 |       "text/plain": [
1413 |        "<matplotlib.axes._subplots.AxesSubplot at 0x1ff00a9a860>"
1414 |       ]
1415 |      },
1416 |      "execution_count": 15,
1417 |      "metadata": {},
1418 |      "output_type": "execute_result"
1419 |     },
1420 |     {
1421 |      "data": {
1422 |       "image/png": "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\n",
1423 |       "text/plain": [
1424 |        "<Figure size 432x288 with 1 Axes>"
1425 |       ]
1426 |      },
1427 |      "metadata": {},
1428 |      "output_type": "display_data"
1429 |     }
1430 |    ],
1431 |    "source": [
1432 |     "df_hr_act = train_df['heartrate'].groupby(train_df['activityID']).mean()\n",
1433 |     "df_hr_act.index = df_hr_act.index.map(activityIDdict)\n",
1434 |     "df_hr_act.plot(kind='bar')"
1435 |    ]
1436 |   },
1437 |   {
1438 |    "cell_type": "markdown",
1439 |    "metadata": {},
1440 |    "source": [
1441 |     "The bar chart shows that Rope Jumping and Running are the most cumbersome activities out of all the activities.\n",
1442 |     "\n",
1443 |     "To check further on our data to see any anormalities, we have to plot a heat map which will show whether our data has correlations inbetween it. All columns will be used in order to understand the extend of problems, if there are any."
1444 |    ]
1445 |   },
1446 |   {
1447 |    "cell_type": "code",
1448 |    "execution_count": 16,
1449 |    "metadata": {},
1450 |    "outputs": [
1451 |     {
1452 |      "data": {
1453 |       "image/png": "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\n",
1454 |       "text/plain": [
1455 |        "<Figure size 1080x720 with 2 Axes>"
1456 |       ]
1457 |      },
1458 |      "metadata": {},
1459 |      "output_type": "display_data"
1460 |     }
1461 |    ],
1462 |    "source": [
1463 |     "from pandas.plotting import scatter_matrix\n",
1464 |     "df_corr = train_df.corr()\n",
1465 |     "df_corr = df_corr.drop(['activityID'], axis = 1)\n",
1466 |     "\n",
1467 |     "f, ax = plt.subplots(figsize=(15, 10))\n",
1468 |     "sns.heatmap(df_corr, mask=np.zeros_like(df_corr, dtype=np.bool), cmap = \"BrBG\",ax=ax)\n",
1469 |     "plt.show()"
1470 |    ]
1471 |   },
1472 |   {
1473 |    "cell_type": "markdown",
1474 |    "metadata": {},
1475 |    "source": [
1476 |     "Our heatmap shows how much statistical similarity there is between our different columns. We can every easily observe that the gyroscopes do not correlate with any of our other data and seem unneeded in this model. \n",
1477 |     "\n",
1478 |     "On the other hand we can understand the correlation between accelerometers of the hand and temperature. The two are strongly correlated on all three instances of hand accelerometers. \n",
1479 |     "\n",
1480 |     "Furthermore the chest Magnetometers seem to be correlated with heart rate and it is very logical as they very close together on the body."
1481 |    ]
1482 |   },
1483 |   {
1484 |    "cell_type": "markdown",
1485 |    "metadata": {},
1486 |    "source": [
1487 |     "## Hypothesis Testing"
1488 |    ]
1489 |   },
1490 |   {
1491 |    "cell_type": "markdown",
1492 |    "metadata": {},
1493 |    "source": [
1494 |     "The most cumbersome activities seem to be running and rope jumping as seen from bar charts plotted above. Therefore our hypothesis testing will be based on these two activities and how their heart rate data correlates with the rest of the activities' heart rates. This will be done by getting the mean heart rate of the two activities and them comparing it to the mean heart rate of all activities."
1495 |    ]
1496 |   },
1497 |   {
1498 |    "cell_type": "markdown",
1499 |    "metadata": {},
1500 |    "source": [
1501 |     "**Null Hypothesis: **\n",
1502 |     "- h0 : The mean heart rate of the cumbersome activities has no mass difference from the mean of all activities\n",
1503 |     "\n",
1504 |     "**Non Null Hypothesis: **\n",
1505 |     "\n",
1506 |     "- h1 : The mean heart rate of the cumbersome activities has mass difference from the mean of all activities"
1507 |    ]
1508 |   },
1509 |   {
1510 |    "cell_type": "code",
1511 |    "execution_count": 17,
1512 |    "metadata": {},
1513 |    "outputs": [],
1514 |    "source": [
1515 |     "running_data = train_df.loc[(train_df[\"activityID\"] == 5)]\n",
1516 |     "ropejumping_data = train_df.loc[(train_df[\"activityID\"] == 24)]\n",
1517 |     "cumbersome_data = running_data + ropejumping_data"
1518 |    ]
1519 |   },
1520 |   {
1521 |    "cell_type": "code",
1522 |    "execution_count": 18,
1523 |    "metadata": {},
1524 |    "outputs": [
1525 |     {
1526 |      "name": "stdout",
1527 |      "output_type": "stream",
1528 |      "text": [
1529 |       "The p_value is  0.0  and h0 is rejected. There is mass difference between the means of cumbersome activities and all activities.\n"
1530 |      ]
1531 |     }
1532 |    ],
1533 |    "source": [
1534 |     "import scipy.stats\n",
1535 |     "\n",
1536 |     "p = train_df['heartrate'].mean() / (running_data['heartrate'].std() / math.sqrt( running_data['heartrate'].count() ))\n",
1537 |     "pValue = 1 - scipy.stats.norm.cdf(p)\n",
1538 |     "\n",
1539 |     "if pValue > 0.1:\n",
1540 |     "    print(\"The p_value is \", pValue, \" and h1 is rejected. There is no mass difference between the means of cumbersome activities and all activities.\")\n",
1541 |     "else:\n",
1542 |     "    print(\"The p_value is \", pValue, \" and h0 is rejected. There is mass difference between the means of cumbersome activities and all activities.\")\n"
1543 |    ]
1544 |   },
1545 |   {
1546 |    "cell_type": "markdown",
1547 |    "metadata": {},
1548 |    "source": [
1549 |     "After making sure that our hypothesis is non null, proving that the mean of the heart rate of the two most cumbersome activities is in fact very different from the mean heart rate of all the activities we can move on into Modelling where we will be looking at different modelling algorithms and will be choosing one as the modelling algorithm that would be used after testing."
1550 |    ]
1551 |   },
1552 |   {
1553 |    "cell_type": "markdown",
1554 |    "metadata": {},
1555 |    "source": [
1556 |     "## Modelling"
1557 |    ]
1558 |   },
1559 |   {
1560 |    "cell_type": "markdown",
1561 |    "metadata": {},
1562 |    "source": [
1563 |     "Some variables have to be dropped which would impact our modelling precision. The variables to be dropped are timestamp and subject_id as they are numeric numbers which would our modelling method would use in its calculations but since their values don't have any meaning, the modelling method used would have noise and predictions of accuracy would be innacurate."
1564 |    ]
1565 |   },
1566 |   {
1567 |    "cell_type": "code",
1568 |    "execution_count": 19,
1569 |    "metadata": {},
1570 |    "outputs": [],
1571 |    "source": [
1572 |     "from sklearn.tree import DecisionTreeClassifier\n",
1573 |     "from sklearn.model_selection import train_test_split,cross_val_score,StratifiedShuffleSplit\n",
1574 |     "from sklearn.metrics import precision_score,recall_score, f1_score, confusion_matrix,roc_auc_score,roc_curve, accuracy_score\n",
1575 |     "from sklearn.preprocessing import StandardScaler, RobustScaler\n",
1576 |     "from sklearn.decomposition import PCA, TruncatedSVD\n",
1577 |     "from sklearn.linear_model import LogisticRegression\n",
1578 |     "from sklearn.neighbors import KNeighborsClassifier\n",
1579 |     "from sklearn.svm import SVC"
1580 |    ]
1581 |   },
1582 |   {
1583 |    "cell_type": "code",
1584 |    "execution_count": 20,
1585 |    "metadata": {},
1586 |    "outputs": [],
1587 |    "source": [
1588 |     "train_df = train_df.drop([\"timestamp\", \"subject_id\"],1)"
1589 |    ]
1590 |   },
1591 |   {
1592 |    "cell_type": "code",
1593 |    "execution_count": 21,
1594 |    "metadata": {},
1595 |    "outputs": [
1596 |     {
1597 |      "data": {
1598 |       "text/html": [
1599 |        "<div>\n",
1600 |        "<style scoped>\n",
1601 |        "    .dataframe tbody tr th:only-of-type {\n",
1602 |        "        vertical-align: middle;\n",
1603 |        "    }\n",
1604 |        "\n",
1605 |        "    .dataframe tbody tr th {\n",
1606 |        "        vertical-align: top;\n",
1607 |        "    }\n",
1608 |        "\n",
1609 |        "    .dataframe thead th {\n",
1610 |        "        text-align: right;\n",
1611 |        "    }\n",
1612 |        "</style>\n",
1613 |        "<table border=\"1\" class=\"dataframe\">\n",
1614 |        "  <thead>\n",
1615 |        "    <tr style=\"text-align: right;\">\n",
1616 |        "      <th></th>\n",
1617 |        "      <th>activityID</th>\n",
1618 |        "      <th>heartrate</th>\n",
1619 |        "      <th>handTemperature</th>\n",
1620 |        "      <th>handAcc16_1</th>\n",
1621 |        "      <th>handAcc16_2</th>\n",
1622 |        "      <th>handAcc16_3</th>\n",
1623 |        "      <th>handAcc6_1</th>\n",
1624 |        "      <th>handAcc6_2</th>\n",
1625 |        "      <th>handAcc6_3</th>\n",
1626 |        "      <th>handGyro1</th>\n",
1627 |        "      <th>...</th>\n",
1628 |        "      <th>ankleAcc16_3</th>\n",
1629 |        "      <th>ankleAcc6_1</th>\n",
1630 |        "      <th>ankleAcc6_2</th>\n",
1631 |        "      <th>ankleAcc6_3</th>\n",
1632 |        "      <th>ankleGyro1</th>\n",
1633 |        "      <th>ankleGyro2</th>\n",
1634 |        "      <th>ankleGyro3</th>\n",
1635 |        "      <th>ankleMagne1</th>\n",
1636 |        "      <th>ankleMagne2</th>\n",
1637 |        "      <th>ankleMagne3</th>\n",
1638 |        "    </tr>\n",
1639 |        "  </thead>\n",
1640 |        "  <tbody>\n",
1641 |        "    <tr>\n",
1642 |        "      <th>312921</th>\n",
1643 |        "      <td>3</td>\n",
1644 |        "      <td>-0.368421</td>\n",
1645 |        "      <td>0.447368</td>\n",
1646 |        "      <td>-0.384512</td>\n",
1647 |        "      <td>0.220069</td>\n",
1648 |        "      <td>-0.572507</td>\n",
1649 |        "      <td>-0.361436</td>\n",
1650 |        "      <td>0.256971</td>\n",
1651 |        "      <td>-0.530520</td>\n",
1652 |        "      <td>-0.012729</td>\n",
1653 |        "      <td>...</td>\n",
1654 |        "      <td>0.321939</td>\n",
1655 |        "      <td>0.088339</td>\n",
1656 |        "      <td>-0.322152</td>\n",
1657 |        "      <td>0.351800</td>\n",
1658 |        "      <td>0.220183</td>\n",
1659 |        "      <td>0.001165</td>\n",
1660 |        "      <td>-0.139354</td>\n",
1661 |        "      <td>0.641700</td>\n",
1662 |        "      <td>-0.582432</td>\n",
1663 |        "      <td>0.658114</td>\n",
1664 |        "    </tr>\n",
1665 |        "    <tr>\n",
1666 |        "      <th>141735</th>\n",
1667 |        "      <td>12</td>\n",
1668 |        "      <td>1.681818</td>\n",
1669 |        "      <td>0.236842</td>\n",
1670 |        "      <td>-0.721708</td>\n",
1671 |        "      <td>-0.103407</td>\n",
1672 |        "      <td>-0.206281</td>\n",
1673 |        "      <td>-0.625117</td>\n",
1674 |        "      <td>-0.108947</td>\n",
1675 |        "      <td>-0.253147</td>\n",
1676 |        "      <td>-0.868631</td>\n",
1677 |        "      <td>...</td>\n",
1678 |        "      <td>0.020813</td>\n",
1679 |        "      <td>0.842160</td>\n",
1680 |        "      <td>-0.269519</td>\n",
1681 |        "      <td>-0.185601</td>\n",
1682 |        "      <td>-0.829881</td>\n",
1683 |        "      <td>3.024896</td>\n",
1684 |        "      <td>-0.265261</td>\n",
1685 |        "      <td>-0.744416</td>\n",
1686 |        "      <td>1.091512</td>\n",
1687 |        "      <td>-0.924482</td>\n",
1688 |        "    </tr>\n",
1689 |        "    <tr>\n",
1690 |        "      <th>1191085</th>\n",
1691 |        "      <td>24</td>\n",
1692 |        "      <td>2.026316</td>\n",
1693 |        "      <td>0.289474</td>\n",
1694 |        "      <td>0.457975</td>\n",
1695 |        "      <td>0.035504</td>\n",
1696 |        "      <td>-2.015047</td>\n",
1697 |        "      <td>0.603512</td>\n",
1698 |        "      <td>0.072534</td>\n",
1699 |        "      <td>-1.452154</td>\n",
1700 |        "      <td>-5.190897</td>\n",
1701 |        "      <td>...</td>\n",
1702 |        "      <td>1.905139</td>\n",
1703 |        "      <td>-7.260794</td>\n",
1704 |        "      <td>-0.946788</td>\n",
1705 |        "      <td>1.482238</td>\n",
1706 |        "      <td>5.524513</td>\n",
1707 |        "      <td>-0.773631</td>\n",
1708 |        "      <td>-1.222988</td>\n",
1709 |        "      <td>-0.447203</td>\n",
1710 |        "      <td>-0.428537</td>\n",
1711 |        "      <td>-0.674781</td>\n",
1712 |        "    </tr>\n",
1713 |        "    <tr>\n",
1714 |        "      <th>1206914</th>\n",
1715 |        "      <td>1</td>\n",
1716 |        "      <td>-1.105263</td>\n",
1717 |        "      <td>0.131579</td>\n",
1718 |        "      <td>1.290689</td>\n",
1719 |        "      <td>-0.693932</td>\n",
1720 |        "      <td>0.901352</td>\n",
1721 |        "      <td>1.320385</td>\n",
1722 |        "      <td>-0.723956</td>\n",
1723 |        "      <td>0.911347</td>\n",
1724 |        "      <td>0.065061</td>\n",
1725 |        "      <td>...</td>\n",
1726 |        "      <td>-0.351884</td>\n",
1727 |        "      <td>-5.128883</td>\n",
1728 |        "      <td>-2.300004</td>\n",
1729 |        "      <td>-0.300736</td>\n",
1730 |        "      <td>-0.037313</td>\n",
1731 |        "      <td>0.027350</td>\n",
1732 |        "      <td>0.037709</td>\n",
1733 |        "      <td>0.700615</td>\n",
1734 |        "      <td>0.847494</td>\n",
1735 |        "      <td>-0.859194</td>\n",
1736 |        "    </tr>\n",
1737 |        "    <tr>\n",
1738 |        "      <th>1710520</th>\n",
1739 |        "      <td>2</td>\n",
1740 |        "      <td>-0.657895</td>\n",
1741 |        "      <td>0.500000</td>\n",
1742 |        "      <td>0.434876</td>\n",
1743 |        "      <td>-2.413007</td>\n",
1744 |        "      <td>-0.473610</td>\n",
1745 |        "      <td>0.449008</td>\n",
1746 |        "      <td>-2.412753</td>\n",
1747 |        "      <td>-0.481267</td>\n",
1748 |        "      <td>-0.015890</td>\n",
1749 |        "      <td>...</td>\n",
1750 |        "      <td>-0.107595</td>\n",
1751 |        "      <td>-0.192439</td>\n",
1752 |        "      <td>0.783229</td>\n",
1753 |        "      <td>-0.057162</td>\n",
1754 |        "      <td>0.016440</td>\n",
1755 |        "      <td>0.068780</td>\n",
1756 |        "      <td>0.058657</td>\n",
1757 |        "      <td>0.611770</td>\n",
1758 |        "      <td>0.497042</td>\n",
1759 |        "      <td>0.263863</td>\n",
1760 |        "    </tr>\n",
1761 |        "  </tbody>\n",
1762 |        "</table>\n",
1763 |        "<p>5 rows × 41 columns</p>\n",
1764 |        "</div>"
1765 |       ],
1766 |       "text/plain": [
1767 |        "         activityID  heartrate  handTemperature  handAcc16_1  handAcc16_2  \\\n",
1768 |        "312921            3  -0.368421         0.447368    -0.384512     0.220069   \n",
1769 |        "141735           12   1.681818         0.236842    -0.721708    -0.103407   \n",
1770 |        "1191085          24   2.026316         0.289474     0.457975     0.035504   \n",
1771 |        "1206914           1  -1.105263         0.131579     1.290689    -0.693932   \n",
1772 |        "1710520           2  -0.657895         0.500000     0.434876    -2.413007   \n",
1773 |        "\n",
1774 |        "         handAcc16_3  handAcc6_1  handAcc6_2  handAcc6_3  handGyro1  \\\n",
1775 |        "312921     -0.572507   -0.361436    0.256971   -0.530520  -0.012729   \n",
1776 |        "141735     -0.206281   -0.625117   -0.108947   -0.253147  -0.868631   \n",
1777 |        "1191085    -2.015047    0.603512    0.072534   -1.452154  -5.190897   \n",
1778 |        "1206914     0.901352    1.320385   -0.723956    0.911347   0.065061   \n",
1779 |        "1710520    -0.473610    0.449008   -2.412753   -0.481267  -0.015890   \n",
1780 |        "\n",
1781 |        "            ...       ankleAcc16_3  ankleAcc6_1  ankleAcc6_2  ankleAcc6_3  \\\n",
1782 |        "312921      ...           0.321939     0.088339    -0.322152     0.351800   \n",
1783 |        "141735      ...           0.020813     0.842160    -0.269519    -0.185601   \n",
1784 |        "1191085     ...           1.905139    -7.260794    -0.946788     1.482238   \n",
1785 |        "1206914     ...          -0.351884    -5.128883    -2.300004    -0.300736   \n",
1786 |        "1710520     ...          -0.107595    -0.192439     0.783229    -0.057162   \n",
1787 |        "\n",
1788 |        "         ankleGyro1  ankleGyro2  ankleGyro3  ankleMagne1  ankleMagne2  \\\n",
1789 |        "312921     0.220183    0.001165   -0.139354     0.641700    -0.582432   \n",
1790 |        "141735    -0.829881    3.024896   -0.265261    -0.744416     1.091512   \n",
1791 |        "1191085    5.524513   -0.773631   -1.222988    -0.447203    -0.428537   \n",
1792 |        "1206914   -0.037313    0.027350    0.037709     0.700615     0.847494   \n",
1793 |        "1710520    0.016440    0.068780    0.058657     0.611770     0.497042   \n",
1794 |        "\n",
1795 |        "         ankleMagne3  \n",
1796 |        "312921      0.658114  \n",
1797 |        "141735     -0.924482  \n",
1798 |        "1191085    -0.674781  \n",
1799 |        "1206914    -0.859194  \n",
1800 |        "1710520     0.263863  \n",
1801 |        "\n",
1802 |        "[5 rows x 41 columns]"
1803 |       ]
1804 |      },
1805 |      "execution_count": 21,
1806 |      "metadata": {},
1807 |      "output_type": "execute_result"
1808 |     }
1809 |    ],
1810 |    "source": [
1811 |     "from sklearn import preprocessing\n",
1812 |     "from sklearn.preprocessing import StandardScaler,RobustScaler\n",
1813 |     "\n",
1814 |     "#apply scaling to all columns except subject and activity \n",
1815 |     "scaler = RobustScaler()\n",
1816 |     "df_scaled = train_df.copy()\n",
1817 |     "df_scaled_test = test_df.copy()\n",
1818 |     "\n",
1819 |     "df_scaled.iloc[:,1:41] = scaler.fit_transform(df_scaled.iloc[:,1:41])\n",
1820 |     "df_scaled_test.iloc[:,1:41] = scaler.fit_transform(df_scaled_test.iloc[:,1:41])\n",
1821 |     "\n",
1822 |     "df_scaled.head()"
1823 |    ]
1824 |   },
1825 |   {
1826 |    "cell_type": "code",
1827 |    "execution_count": 22,
1828 |    "metadata": {},
1829 |    "outputs": [],
1830 |    "source": [
1831 |     "X_train = df_scaled.drop('activityID', axis=1).values\n",
1832 |     "y_train = df_scaled['activityID'].values\n",
1833 |     "\n",
1834 |     "# Test Dataset\n",
1835 |     "X_test = df_scaled.drop('activityID', axis=1).values\n",
1836 |     "y_test = df_scaled['activityID'].values"
1837 |    ]
1838 |   },
1839 |   {
1840 |    "cell_type": "markdown",
1841 |    "metadata": {},
1842 |    "source": [
1843 |     "### Dimensionality reduction using Principal Component Analysis(PCA)\n",
1844 |     "\n",
1845 |     "\n",
1846 |     "Usually 90-98% of the variance will explain our data really well. So by plotting the variance ratio aginst the number of componments we could see how many of those we could use. As we see from the graph below 15 componments fall around to 94% of the variance. "
1847 |    ]
1848 |   },
1849 |   {
1850 |    "cell_type": "code",
1851 |    "execution_count": 23,
1852 |    "metadata": {},
1853 |    "outputs": [
1854 |     {
1855 |      "data": {
1856 |       "image/png": "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\n",
1857 |       "text/plain": [
1858 |        "<Figure size 432x288 with 1 Axes>"
1859 |       ]
1860 |      },
1861 |      "metadata": {},
1862 |      "output_type": "display_data"
1863 |     }
1864 |    ],
1865 |    "source": [
1866 |     "from sklearn.decomposition import PCA\n",
1867 |     "pca = PCA()\n",
1868 |     "pca.fit(X_train)\n",
1869 |     "var= pca.explained_variance_ratio_\n",
1870 |     "var1=np.cumsum(np.round(pca.explained_variance_ratio_, decimals=4)*100)\n",
1871 |     "\n",
1872 |     "plt.title(\"PCA Variance against num of Componmnets\")\n",
1873 |     "plt.ylabel(\"Variance %\")\n",
1874 |     "plt.xlabel(\"Number of componments\")\n",
1875 |     "l = plt.axhline(94, color=\"red\")\n",
1876 |     "\n",
1877 |     "plt.plot(var1)\n",
1878 |     "plt.grid()\n"
1879 |    ]
1880 |   },
1881 |   {
1882 |    "cell_type": "code",
1883 |    "execution_count": null,
1884 |    "metadata": {},
1885 |    "outputs": [],
1886 |    "source": [
1887 |     "pca = PCA(n_components=17)\n",
1888 |     "X_train=pca.fit_transform(X_train)\n",
1889 |     "X_test=pca.fit_transform(X_test)"
1890 |    ]
1891 |   },
1892 |   {
1893 |    "cell_type": "markdown",
1894 |    "metadata": {},
1895 |    "source": [
1896 |     "### Choosing the best model\n",
1897 |     "In this section we will train two models and check their performance.\n",
1898 |     "\n",
1899 |     "We are going to focus on two modelling algorithms, Random Forest and Logistic Regression.\n",
1900 |     "\n",
1901 |     "**Random Forest** algorithm can be used for both classification and regression which makes it a very versatile modelling algorithm. As the name implies, Random Forest is a forest of trees, decision trees which are randomly populating the forest. The algorithms creates and combines decision trees together, the more trees in the forest, the better the accuracy of its predictions will be.\n",
1902 |     "\n",
1903 |     "Random Forest algorithms are good because of the high accuracy they provide as well as that they are flexible i.e. can work for both classification and regression modelling. In addition, Random Forest facilitates for good performance of high dimensionality datasets which our dataset has. \n",
1904 |     "\n",
1905 |     "**Logistic Regression** algorithm is a simple algorithm that can be used for binary/multivariate classification tasks. The result of it is a probability that a data point is part of a class."
1906 |    ]
1907 |   },
1908 |   {
1909 |    "cell_type": "code",
1910 |    "execution_count": null,
1911 |    "metadata": {},
1912 |    "outputs": [],
1913 |    "source": [
1914 |     "def get_metrics (y_true,y_pred):\n",
1915 |     "    acc = accuracy_score(y_true, y_pred)\n",
1916 |     "    err = 1-acc\n",
1917 |     "    p = precision_score(y_true, y_pred,average=None).mean()\n",
1918 |     "    r = recall_score(y_true, y_pred, average=None).mean()\n",
1919 |     "    f1 = f1_score(y_true, y_pred, average=None).mean()\n",
1920 |     "   \n",
1921 |     "    print(\"Accuracy: \",acc)\n",
1922 |     "    print(\"Error: \",acc)\n",
1923 |     "    print(\"Precision\", p)\n",
1924 |     "    print(\"Recall\", r)\n",
1925 |     "    print(\"F1\", f1)\n",
1926 |     "    #print(\"Accuracy: %.3f \\nError: %.3f \\nPrecesion: %.3f \\nRecall: %.3f \\nF1:%.3f \\n\"% (acc,err,p,r,f1) )\n",
1927 |     "    \n",
1928 |     "\n",
1929 |     "log_reg = LogisticRegression()\n",
1930 |     "log_reg.fit(X=X_train, y=y_train )\n",
1931 |     "y_pred_lr = log_reg.predict(X_test)\n",
1932 |     "get_metrics(y_test, y_pred_lr)"
1933 |    ]
1934 |   },
1935 |   {
1936 |    "cell_type": "code",
1937 |    "execution_count": null,
1938 |    "metadata": {},
1939 |    "outputs": [],
1940 |    "source": [
1941 |     "from sklearn.ensemble import RandomForestClassifier\n",
1942 |     "rfc = RandomForestClassifier(n_jobs =4)\n",
1943 |     "rfc.fit(X_train,y_train)\n",
1944 |     "y_pred_rf = rfc.predict(X_test)\n",
1945 |     "get_metrics(y_test,y_pred_rf)"
1946 |    ]
1947 |   },
1948 |   {
1949 |    "cell_type": "markdown",
1950 |    "metadata": {},
1951 |    "source": [
1952 |     "###  Cross validation"
1953 |    ]
1954 |   },
1955 |   {
1956 |    "cell_type": "markdown",
1957 |    "metadata": {},
1958 |    "source": [
1959 |     "Even though the above models seem to perform really good, the metrics used for that do not represent the real score since the models were train on a specific part of the dataset. By using cross validation, we could k=10 number of folds, which in few words, will generate 10 different samples. By doing that we will get 10 different metrics values. The mean value of these metrics will show a better representation of our model's performance"
1960 |    ]
1961 |   },
1962 |   {
1963 |    "cell_type": "code",
1964 |    "execution_count": null,
1965 |    "metadata": {},
1966 |    "outputs": [],
1967 |    "source": [
1968 |     "from sklearn.model_selection import cross_val_score\n",
1969 |     "from sklearn.model_selection import cross_val_predict\n",
1970 |     "\n",
1971 |     "classifiers = [LogisticRegression(),\n",
1972 |     "               RandomForestClassifier(n_jobs =4)]\n",
1973 |     "\n",
1974 |     "score_lst = []\n",
1975 |     "for cls in classifiers:\n",
1976 |     "    accs  = accuracy_score(y_train, cross_val_predict(cls,X_train,y_train,cv = 10))  #cross validate the accurancy \n",
1977 |     "    scores = cross_val_score(cls,X_train,y_train,scoring = \"neg_mean_squared_error\",cv= 10)#calcuate the error\n",
1978 |     "    score = np.sqrt(-scores)\n",
1979 |     "    f1 = cross_val_score(cls,X_test, y_test,scoring = \"f1_macro\", cv =10)\n",
1980 |     "    \n",
1981 |     "    \n",
1982 |     "    #p = cross_val_score(cls,X_test, y_test,scoring = \"average_precision\", cv =10)\n",
1983 |     "#     r = recall_score(y_true, y_pred, average=None).mean()\n",
1984 |     "#     f1 = f1_score(y_true, y_pred, average=None).mean()\n",
1985 |     "    score_lst.append([cls.__class__.__name__,accs,score.mean(), f1.mean()])\n",
1986 |     "\n",
1987 |     "df_scores=pd.DataFrame(columns = [\"Claasifier\",\"Accurancy\",\"MSE\",\"F1\"],data =  score_lst)\n",
1988 |     "display(df_scores)"
1989 |    ]
1990 |   },
1991 |   {
1992 |    "cell_type": "markdown",
1993 |    "metadata": {},
1994 |    "source": [
1995 |     "As this took a lot of time to compute the results of the above cells, it is uploaded without the outputs but work and should be run by the examiner as the computer that I am using has very few processing capabilities."
1996 |    ]
1997 |   },
1998 |   {
1999 |    "cell_type": "markdown",
2000 |    "metadata": {},
2001 |    "source": [
2002 |     "## References\n",
2003 |     "\n",
2004 |     "Archive.ics.uci.edu. (2012). UCI Machine Learning Repository: PAMAP2 Physical Activity Monitoring Data Set. [online] Available at: http://archive.ics.uci.edu/ml/datasets/pamap2+physical+activity+monitoring [Accessed 15 Dec. 2018].\n",
2005 |     "\n",
2006 |     "Chandrayan, P. (2017). Machine Learning Part 3 : Logistic Regression – Towards Data Science. [online] Towards Data Science. Available at: https://towardsdatascience.com/machine-learning-part-3-logistics-regression-9d890928680f [Accessed 2 Jan. 2019].\n",
2007 |     "\n",
2008 |     "DeZyre. (n.d.). Principal Component Analysis Tutorial. [online] Available at: https://www.dezyre.com/data-science-in-python-tutorial/principal-component-analysis-tutorial [Accessed 30 Dec. 2018].\n",
2009 |     "\n",
2010 |     "Donges, N. (2018). The Random Forest Algorithm – Towards Data Science. [online] Towards Data Science. Available at: https://towardsdatascience.com/the-random-forest-algorithm-d457d499ffcd [Accessed 3 Jan. 2019].\n",
2011 |     "\n",
2012 |     "PAMAP2_Dataset: Physical Activity Monitoring. (n.d.). [ebook] Available at: http://archive.ics.uci.edu/ml/machine-learning-databases/00231/readme.pdf [Accessed 15 Dec. 2018].\n",
2013 |     "\n",
2014 |     "Scikit-learn.org. (n.d.). Robust Scaling on Toy Data — scikit-learn 0.18.2 documentation. [online] Available at: https://scikit-learn.org/0.18/auto_examples/preprocessing/plot_robust_scaling.html [Accessed 2 Jan. 2019]."
2015 |    ]
2016 |   },
2017 |   {
2018 |    "cell_type": "code",
2019 |    "execution_count": null,
2020 |    "metadata": {},
2021 |    "outputs": [],
2022 |    "source": []
2023 |   }
2024 |  ],
2025 |  "metadata": {
2026 |   "kernelspec": {
2027 |    "display_name": "Python 3",
2028 |    "language": "python",
2029 |    "name": "python3"
2030 |   },
2031 |   "language_info": {
2032 |    "codemirror_mode": {
2033 |     "name": "ipython",
2034 |     "version": 3
2035 |    },
2036 |    "file_extension": ".py",
2037 |    "mimetype": "text/x-python",
2038 |    "name": "python",
2039 |    "nbconvert_exporter": "python",
2040 |    "pygments_lexer": "ipython3",
2041 |    "version": "3.6.5"
2042 |   }
2043 |  },
2044 |  "nbformat": 4,
2045 |  "nbformat_minor": 2
2046 | }
2047 | 


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