classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Author: Corentin Llorca"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Classification using the Telco-churn dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import warnings \n",
    "warnings.simplefilter('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We start by loading the data. The feature 'SeniorCitizen' will be read as a number by pandas, so we enforce it as categorical. We also notice that TotalCharges is read as an object, so we enforce it as a float."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"data/class.csv\", dtype={'SeniorCitizen': object}, na_values=[' '])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing the distributions of features and target"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here are the features: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAoAAAAGoCAYAAADW2lTlAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzs3XuYZVV95//3JyCKEmku2oM0sXFCjBdGIz1K4oxpxCigEf2NJhBGQElIZjDRSCY2MRNMjDMkERWNo2mFCIbQENTAKIkStMYxIygoERANDbbQ0NIi0NrgrfX7+2PvktPV1V116lSdc6r2+/U856m9115n7+86l13fs9a+pKqQJElSd/zEqAOQJEnScJkASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmghi5JJfnpnSw7Ocmn52k7b0zyN/OxLklaCEke0e4TV8yy/lFJ1i90XFr6TAA7LsmGJN9Psv+U8uvbndLKAdc/keTXB1nHDOv/tSTXJtmaZFOSf0jyHxZqe5KWvnZ/Mvn4UZLv9MyfMMNzB07QkvxCko8n2ZLkm0munmm7Ur9MAAXwVeD4yZkkhwJ7ji6c2UnyOuDtwP8AlgM/Bfwv4NgF2Nbu871OSeOpqvaafAC3A7/cU3bhQm47yWrg48A/AgcD+wO/A7xwAbblfq3DTAAF8AHgxJ75k4ALJmeS7J3kgiTfSPK1JH+Y5CfaZScn+XSStyS5L8lXkxzdLnsz8B+Bv2x/Of9lzzael+SW9jnvSpKpQbXlZ08p+99JXptkb+BPgNOq6kNV9UBV/aCq/ndV/beep+zRxv7tJDclWdWzrjVJbm2XfSnJS3uWnZzkn5O8Lcm9wBuT7Jbk7CT3tO18ddtLunvP63Ru2xN5Z5I/TbJbu+ynk/yf9hf9PUku7u8tkjQukuzZ7p82JdmY5C+SPCzJfsCHgSf09Bjul+TZSa5pv/93tfuVnSVfbwH+qqreWlX3VuOzVfVrU2L4g3affGdv72CSlyb5lyTfavfXf9Cz7GeTbEvyG0nuAK5oy389ye3t+n4/ydcnR1La/d5/T3Jbu++6MMmydtmjkqxLcm+S+9s27jOvL7YWjAmgAK4GHp3kSW3C8qtA77Fz7wT2Bp4A/CJNsvjKnuXPAr5C80v1z4Fzk6Sq3gD8X+DV7S/nV/c850XAvweeBvwK8IJp4jofOL4n2dwfOBK4CPh54BE0O9tdeTGwDlgGXA70JqG30iSoewN/DPxNkgOmtOs24LHAm4HfAI4Gng48A3jJNPFuA34a+Dng+cDk8PebaH7V7wOsoHlNJS1Ofwz8O+BQ4DBgNfD7VfVN4KXAbT09ht8EfgC8GtiXZp/zyzy0b/ixNrE6DLh0hu0/HgjwuHa970myV7vsW8Cv0ezzXgr8XpKjep67G82+7YnAsUmeDryVZj+8on30HhL032j2Zf+hXfYD4G3tsl8HdgcObJ/zauD7M8SuMWECqEmTvYC/BHwZuLMtn0wIz6iqb1fVBuBs4BU9z/1aVb23qn5IkwQdQDMkuytnVdX9VXU78EmapGo7VfVZYAtN0gdwHDBRVXcD+wH3VNW2Gbbz6aq6oo3tAzQJ5+T6/66q7qqqH1XVxcAtwDN7nntXVb2zqrZV1XdodpDnVNXGqroPOGuyYpLlNMnha9veyM00O8nj2io/oNlpP66qvltV83Kii6SROAE4s6ruafdHf8r2+8TttD14n6uqH1bVrcD7aH5MT7Vf+3fTDNt/EPif7ajHh4Gi+eFJVV1VVTe1+7XPA5dMs60/qqoHe/ZrH6yqq6vqe8Afsn1u8JvAmnZf+V2a5PdX21GbHwCPAf5tu5/8XFU9MEPsGhMmgJr0AZpfjSfTM/xL86tuD+BrPWVfo/nFN+nrkxNV9WA7uRe79vWe6Qd3Uf984D+30/+5jRPgm8D+sziGZep2HtEzZHtimpNd7k9yP/BUtv/le8eUdT1uSlnv9OOBhwGbetb3VzS9hwC/T/OL/bPtUPSrZohb0hhqE59/w673iVOf8+Q0J6jdneRbwB+x/b5m0jfbvwdMs6zXN6rqRz3zP96HtsPN/6cdzt1Cs0/v3daPququnvnt9mtV9S2aH96TbT0IuKJnv/YFmtxhP+Bc4P8Al7ZD4f9j8rAXjT8TQAFQVV+jORnkGOBDPYvu4aHeq0k/xUM9hDOuesDQ/oZmmOJpwJOAv2/LPwN8lx2HYWclyeOB99IMWexXVcuAG2mStElTY99EMwQy6aCe6TuA7wH7V9Wy9vHoqnoKQFV9vap+o6oeR/OL+n9lJ5fCkTS+qqpofljubJ843T7vvcDnaXrKHk1z/PIOxz1X1f3AdcB/GiDES4CLgYOqam/g/fSxX0vyaJrDYibbeifw3J792rKqekTb+/m9qvqjqvpZ4DnAy3lo1ENjzgRQvU6h+aL3duH/kGaH8uYkP9kmTq9j+2MEd+VummMH56SqNgKfo+n5+2A7ZEFVbaH5Ff2uJC9J8sj2IOyjk/z5LFb9KJod4TcAkrySpgdwVy4BXpPkwPZYndf3xLmJ5hi/s5M8OslPJPm3SX6xXf/L89B1vu5rt/3DWb0IksbNRcCZ7QkejwXewEP7xLuBx/Yckwfwk8CWqtqa5Ck0xxPvzO8Bv5XmZLd90zgss7imadtjtxfwzar6bpJfoEnKduUS4D8l+fdJ9qBJTnt7F98DnJXkoHYbj03yy+3089rezZ+gOfZwG+7XFg0TQP1YVd1aVddOs+i3gQdoToj4NPC3wHmzXO05wMvSnO37jjmGdj7NwdYf6C2sqrfSJKN/SJPI3UHTo/f3U1cwVVV9ieZYxs/Q7LAPBf55hqe9lybJ+yLNMMgVbL/DO5FmuPxLNEnepTw0lPPvgWuSbKU5GeU1VfXVmeKUNJb+iOZ7fhNwPc2+Y/KH57/QfMe/1g6b7gv8LvDr7ff/XTQ9dNOqqgmaky6OATbQjML8JfDRmYJqe+x+C3hLkm/THHrydzM85ws0J3p8mKa3bxPNEPD32ip/DvwT8Il2nf+P5iQ4aIa9LwO+TTOCcgVNQqlFIM3nRRpfSZ5D8+t65ZTjXkYqzeVu3lNVj5+xsiQtAu1lXO6lOWFtppNRtIjZA6ixluRhwGuA9406+Utz7a9jkuye5EDgTGa+DI0kjbUkL273b3vRXBLmGpO/pc8EUGMryZOA+2mGUd8+4nCgOZD6j2mGd78A3EwzFCRJi9nLaU5s2UgzrOtt5zrAIWBJkqSOsQdQkiSpY8b6RtD7779/rVy5cqB1PPDAAzzqUY+an4BGZCm0AZZGO2zDeJhtG6677rp7quoxQwhpXvW771sK7+lMbOPSYBsX3mz3e2OdAK5cuZJrr53uqiSzNzExwerVq+cnoBFZCm2ApdEO2zAeZtuGJF+bsdIY6nfftxTe05nYxqXBNi682e73HAKWJEnqmBkTwCTnJdmc5Maesr9I8uUkX0zy4fauCJPLzkiyPslXkrygp/yotmx9kjXz3xRJkiTNxmx6AN8PHDWl7ErgqVX174B/Bc6A5obXNPcBfEr7nP+VZLf25tDvAo4Gngwc39aVJEnSkM2YAFbVp2iuCt5b9vGq2tbOXs1DN5I+FljX3iD6q8B64JntY31V3VZV3wfWtXUlSZI0ZPNxDOCrgH9opw+kuR/rpMmLSu6sXJIkSUM20FnASd4AbAMunCyaploxfaI57RWok5wKnAqwfPlyJiYmBgmRrVu3DryOUVsKbYCl0Q7bMB6WQhskaZTmnAAmOQl4EXBkPXQ7kY3AQT3VVgB3tdM7K99OVa0F1gKsWrWq+j2VeuWaj243f/qhP+TsTz+w0/obznphX+sfhVGfUj5flkI7bMN4WAptmE833LmFk6fs+2ayGPZ9khbOnIaAkxwFvB54cVU92LPocuC4JA9PcjBwCPBZ4HPAIUkOTrIHzYkilw8WuiRJkuZixh7AJBcBq4H9k2wEzqQ56/fhwJVJAK6uqt+qqpuSXAJ8iWZo+LSq+mG7nlcDHwN2A86rqpsWoD2SJEmawYwJYFUdP03xubuo/2bgzdOUXwFc0Vd0kiRJmnfeCUSSJKljTAAlSZI6xgRQkiSpY0wAJUmSOsYEUJKmkeS8JJuT3NhTdnGS69vHhiTXt+Urk3ynZ9l7ep5zWJIbkqxP8o60l06QpFEa6E4gkrSEvR/4S+CCyYKq+tXJ6SRnA1t66t9aVU+fZj3vprm70dU0V0I4iodunylJI2EPoCRNo6o+Bdw73bK2F+9XgIt2tY4kBwCPrqrPtHdMugB4yXzHKkn9sgdQkvr3H4G7q+qWnrKDk3wB+Bbwh1X1f4EDaW6ROWljW7aDQe6DvnxPOP3QbX01YLHdS7kL93+2jUvDYmmjCaAk9e94tu/92wT8VFV9M8lhwN8neQow3fF+NU3ZQPdBf+eFl3H2Df3tzjecMPv1j4Mu3P/ZNi4Ni6WNJoCS1IckuwP/H3DYZFlVfQ/4Xjt9XZJbgZ+h6fFb0fP0FcBdw4tWkqbnMYCS1J/nAV+uqh8P7SZ5TJLd2uknAIcAt1XVJuDbSQ5vjxs8EbhsFEFLUi8TQEmaRpKLgM8AT0yyMckp7aLj2PHkj+cAX0zyL8ClwG9V1eQJJP8FeB+wHrgVzwCWNAYcApakaVTV8TspP3masg8CH9xJ/WuBp85rcJI0IHsAJUmSOsYEUJIkqWNMACVJkjrGBFCSJKljTAAlSZI6ZsYEMMl5STYnubGnbN8kVya5pf27T1ueJO9Isj7JF5M8o+c5J7X1b0ly0sI0R5IkSTOZTQ/g+4GjppStAa6qqkOAq9p5gKNpLoB6CM09Ld8NTcIInAk8C3gmcOZk0ihJkqThmjEBrKpPAfdOKT4WOL+dPh94SU/5BdW4GliW5ADgBcCVVXVvVd0HXMmOSaUkSZKGYK4Xgl7e3uKIqtqU5LFt+YHAHT31NrZlOyvfQZJTaXoPWb58ORMTE30Fdvqh27YPdM8dy3r1u/5R2Lp166KIcyZLoR22YTwshTZI0ijN951AMk1Z7aJ8x8KqtcBagFWrVtXq1av7CuDkNR/dbv70Q7dx9g07b+aGE/pb/yhMTEzQ7+swjpZCO2zDeFgKbZCkUZrrWcB3t0O7tH83t+UbgYN66q0A7tpFuSRJkoZsrgng5cDkmbwnAZf1lJ/Yng18OLClHSr+GPD8JPu0J388vy2TJEnSkM3mMjAXAZ8BnphkY5JTgLOAX0pyC/BL7TzAFcBtwHrgvcB/Baiqe4E3AZ9rH3/SlknSWNrJJbDemOTOJNe3j2N6lp3RXgLrK0le0FN+VFu2PsmaqduRpFGY8RjAqjp+J4uOnKZuAaftZD3nAef1FZ0kjc77gb8ELphS/raqektvQZInA8cBTwEeB/xTkp9pF7+L5ofyRuBzSS6vqi8tZOCSNJP5PglEkpaEqvpUkpWzrH4ssK6qvgd8Ncl6mmueAqyvqtsAkqxr65oAShopbwUnSf15dXuno/N6Lmg/8CWwJGmY7AGUpNl7N83xzNX+PRt4FTu/1NV0P7KnvQTWINdAnel6p9NZbNdR7MK1H23j0rBY2mgCKEmzVFV3T04neS/wkXZ2V5e6mtUlsAa5Buo7L7xsl9c7nc5iuAZqry5c+9E2Lg2LpY0OAUvSLE1e/7T1UmDyDOHLgeOSPDzJwTT3Q/8szVUPDklycJI9aE4UuXyYMUvSdOwBlKRptJfAWg3sn2QjcCawOsnTaYZxNwC/CVBVNyW5hObkjm3AaVX1w3Y9r6a57uluwHlVddOQmyJJOzABlKRp7OQSWOfuov6bgTdPU34FzTVSJWlsOAQsSZLUMSaAkiRJHWMCKEmS1DEmgJIkSR1jAihJktQxJoCSJEkdYwIoSZLUMSaAkiRJHWMCKEmS1DEmgJIkSR1jAihJktQxAyWASX43yU1JbkxyUZJHJDk4yTVJbklycZI92roPb+fXt8tXzkcDJEmS1J85J4BJDgR+B1hVVU8FdgOOA/4MeFtVHQLcB5zSPuUU4L6q+mngbW09SRpLSc5LsjnJjT1lf5Hky0m+mOTDSZa15SuTfCfJ9e3jPT3POSzJDe2P33ckySjaI0m9Bh0C3h3YM8nuwCOBTcBzgUvb5ecDL2mnj23naZcf6Y5Q0hh7P3DUlLIrgadW1b8D/hU4o2fZrVX19PbxWz3l7wZOBQ5pH1PXKUlDt/tcn1hVdyZ5C3A78B3g48B1wP1Vta2tthE4sJ0+ELijfe62JFuA/YB7eteb5FSanSXLly9nYmKir7hOP3TbdvPL99yxrFe/6x+FrVu3Loo4Z7IU2mEbxsMw2lBVn5p6qEpVfbxn9mrgZbtaR5IDgEdX1Wfa+QtofhT/w7wGK0l9mnMCmGQfml69g4H7gb8Djp6mak0+ZRfLHiqoWgusBVi1alWtXr26r7hOXvPR7eZPP3QbZ9+w82ZuOKG/9Y/CxMQE/b4O42gptMM2jIcxacOrgIt75g9O8gXgW8AfVtX/pfnhu7GnTu+PYkkamTkngMDzgK9W1TcAknwI+AVgWZLd217AFcBdbf2NwEHAxnbIeG/g3gG2L0kjkeQNwDbgwrZoE/BTVfXNJIcBf5/kKczyh2+7zjmPfsw00jGdxdYLvBR6rmdiG5eGxdLGQRLA24HDkzySZgj4SOBa4JM0wyLrgJOAy9r6l7fzn2mXf6Kqpt0RStK4SnIS8CLgyMl9WFV9D/heO31dkluBn6H54bui5+m9P4q3M8joxzsvvGyXIx3TWQyjH73GpNd3QdnGpWGxtHHOJ4FU1TU0J3N8HrihXdda4PXA65KspznG79z2KecC+7XlrwPWDBC3JA1dkqNo9nEvrqoHe8ofk2S3dvoJNCd73FZVm4BvJzm8PentRB76USxJIzNIDyBVdSZw5pTi24BnTlP3u8DLB9meJA1LkouA1cD+STbS7OvOAB4OXNlexODq9ozf5wB/kmQb8EPgt6pq8hCX/0JzRvGeNCd/eAKIpJEbKAGUpKWqqo6fpvjcacqoqg8CH9zJsmuBp85jaJI0MG8FJ0mS1DEmgJIkSR1jAihJktQxJoCSJEkdYwIoSZLUMSaAkiRJHWMCKEmS1DEmgJIkSR1jAihJktQxJoCSJEkdYwIoSZLUMSaAkiRJHWMCKEmS1DEmgJI0jSTnJdmc5Maesn2TXJnklvbvPm15krwjyfokX0zyjJ7nnNTWvyXJSaNoiyRNZQIoSdN7P3DUlLI1wFVVdQhwVTsPcDRwSPs4FXg3NAkjcCbwLOCZwJmTSaMkjZIJoCRNo6o+Bdw7pfhY4Px2+nzgJT3lF1TjamBZkgOAFwBXVtW9VXUfcCU7JpWSNHS7jzoASVpEllfVJoCq2pTksW35gcAdPfU2tmU7K99BklNpeg9Zvnw5ExMTsw9qTzj90G2zrg/0tf5xsHXr1kUXc79s49KwWNo4UAKYZBnwPuCpQAGvAr4CXAysBDYAv1JV9yUJcA5wDPAgcHJVfX6Q7UvSmMg0ZbWL8h0Lq9YCawFWrVpVq1evnvXG33nhZZx9Q3+78w0nzH7942BiYoJ+XpPFyDYuDYuljYMOAZ8D/GNV/SzwNOBm+jxGRpIWkbvboV3av5vb8o3AQT31VgB37aJckkZqzglgkkcDzwHOBaiq71fV/fR/jIwkLRaXA5Nn8p4EXNZTfmJ7NvDhwJZ2qPhjwPOT7NOe/PH8tkySRmqQIeAnAN8A/jrJ04DrgNfQ/zEym3pXOshxMLDjcTAzHRuzGMbpF8vxBDNZCu2wDeNhGG1IchGwGtg/yUaas3nPAi5JcgpwO/DytvoVNIe3rKc5xOWVAFV1b5I3AZ9r6/1JVU09sUSShm6QBHB34BnAb1fVNUnO4aHh3unM6liYQY6DATh5zUe3mz/90G27PDZmMRwHs1iOJ5jJUmiHbRgPw2hDVR2/k0VHTlO3gNN2sp7zgPPmMTRJGtggxwBuBDZW1TXt/KU0CWG/x8hIkiRpiOacAFbV14E7kjyxLToS+BL9HyMjSZKkIRr0OoC/DVyYZA/gNprjXn6CPo6RkSRJ0nANlABW1fXAqmkW9XWMjCRJkobHW8FJkiR1jAmgJElSx5gASpIkdYwJoCRJUscMehawJGkRWjnlovkz2XDWCxcoEkmjYA+gJElSx5gASpIkdYxDwNIuTB0mO/3QbTvcb3oqh8okSePOHkBJ6kOSJya5vufxrSSvTfLGJHf2lB/T85wzkqxP8pUkLxhl/JIE9gBKUl+q6ivA0wGS7AbcCXyY5vaWb6uqt/TWT/Jk4DjgKcDjgH9K8jNV9cOhBi5JPewBlKS5OxK4taq+tos6xwLrqup7VfVVmvuhP3Mo0UnSTtgDKElzdxxwUc/8q5OcCFwLnF5V9wEHAlf31NnYlm0nyanAqQDLly9nYmJi1kEs37M5PnUh9RPPQti6devIY1hotnFpWCxtNAGUpDlIsgfwYuCMtujdwJuAav+eDbwKyDRPrx0KqtYCawFWrVpVq1evnnUs77zwMs6+YWF35xtOWL2g65/JxMQE/bwmi5FtXBoWSxtNANUp/V78VtqFo4HPV9XdAJN/AZK8F/hIO7sROKjneSuAu4YVpCRNxwRQmmfeYaEzjqdn+DfJAVW1qZ19KXBjO3058LdJ3kpzEsghwGeHGagkTWUCKI2YCePik+SRwC8Bv9lT/OdJnk4zvLthcllV3ZTkEuBLwDbgNM8AljRqJoBatBzO1ahU1YPAflPKXrGL+m8G3rzQcS2kuXzf/LEija+BE8D2OljXAndW1YuSHAysA/YFPg+8oqq+n+ThwAXAYcA3gV+tqg2Dbl/qGv8RS5IGNR89gK8BbgYe3c7/Gc3FUNcleQ9wCs3ZcacA91XVTyc5rq33q/Ow/YH4z1SSJHXNQBeCTrICeCHwvnY+wHOBS9sq5wMvaaePbedplx/Z1pckSdIQDXonkLcDvw/8qJ3fD7i/qiavSNp7wdMDgTsA2uVbmHIMjSRJkhbenIeAk7wI2FxV1yVZPVk8TdWaxbLe9c75aviw49XwF+IK+cO+wvdiuar4TOa7HQt954PpDOOOCwuh93VfCp+npdAGSRqlQY4BfDbw4iTHAI+gOQbw7cCyJLu3vXy9FzydvBjqxiS7A3sD905d6SBXwwc4ecoxfacfum3er5A/7CviL5aris9kvtsx9b0ehoX4PA3FDQ/8ePL0Q3/I2Z9+YBeVx/8416XynZCkUZnzEHBVnVFVK6pqJc39MD9RVScAnwRe1lY7Cbisnb68nadd/omq2qEHUJIkSQtr0GMAp/N64HVJ1tMc43duW34usF9b/jpgzQJsW5IkSTOYl7GsqpoAJtrp24BnTlPnu8DL52N7khaWdyeRpKVtIXoAJUmSNMZMACVJkjrGBFCSJKljFuH1LCRptJJsAL4N/BDYVlWrkuwLXAysBDYAv1JV97V3PDoHOAZ4EDi5qj4/iriHzWNJpfFlAqhZ8Z7J0g6OqKp7eubXAFdV1VlJ1rTzrweOBg5pH8+iuTf6s4YdrCT1cghYkuZH7/3Op94H/YJqXE1zsfwDRhGgJE2yB1CS+lfAx5MU8FftHYyWV9UmgKralOSxbd0f3we9NXmP9E29KxzkNpiL9RaFU+2qzV24/Z9tXBoWSxtNADtqLkO6kn7s2VV1V5vkXZnky7uoO6v7oA9yG8x3XnjZ4rxF4RS7us1mF27/ZxuXhsXSRoeAJalPVXVX+3cz8GGai9/fPTm02/7d3FafvA/6pN57pEvSSJgASlIfkjwqyU9OTgPPB25k+/udT70P+olpHA5smRwqlqRRWfxjBpI0XMuBDzdXd2F34G+r6h+TfA64JMkpwO08dOvLK2guAbOe5jIwrxx+yJK0PRNASepDe7/zp01T/k3gyGnKCzhtCKFJ0qw5BCxJktQx9gAuEZ7VK0mSZsseQEmSpI6xB1ALZmqv5OmHbuPkXfRUeus4SZKGwx5ASZKkjrEHUJI0FnZ1LPN0IwiOGkhzN+cEMMlBwAXAvwF+BKytqnOS7AtcDKwENgC/UlX3pblo1jk018N6EDi5qj4/WPhaSjyRRZKk4RhkCHgbcHpVPQk4HDgtyZOBNcBVVXUIcFU7D3A0cEj7OBV49wDbliRJ0hzNOQGsqk2TPXhV9W3gZuBA4Fjg/Lba+cBL2uljgQuqcTWwbPK+mZIkSRqeeTkGMMlK4OeAa4Dlk/e5rKpNSR7bVjsQuKPnaRvbsu3uiZnkVJoeQpYvX87ExERfsZx+6Lbt5pfvuWPZoPqNaVBbt26dcZvz3caFsBDvxbDZhumN43dCkrRzAyeASfYCPgi8tqq+1d4fc9qq05TVDgVVa4G1AKtWrarVq1f3Fc/Ug4RPP3QbZ98wv+e6bDhh9byubyYTExPM9Drs6vIq42Ih3othsw07ccMDfT9lkAP4Z/OdkCTt3ED/BZI8jCb5u7CqPtQW353kgLb37wBgc1u+ETio5+krgLsG2b4kqbvmcuKYZw5LjTkfA9ie1XsucHNVvbVn0eXASe30ScBlPeUnpnE4sGVyqFiSFoskByX5ZJKbk9yU5DVt+RuT3Jnk+vZxTM9zzkiyPslXkrxgdNFLUmOQHsBnA68AbkhyfVv2B8BZwCVJTgFuB17eLruC5hIw62kuA/PKAbYtSaMyeQWEzyf5SeC6JFe2y95WVW/prdxeHeE44CnA44B/SvIzVfXDoUYtST3mnABW1aeZ/rg+gCOnqV/AaXPdniSNg3bkYvJEt28nmbwCws4cC6yrqu8BX02yHngm8JkFD1Y76HfY2CFjLVWL+2h2SRqhKVdAeDbw6iQnAtfS9BLeR5McXt3ztMkrIExd15yvgLAUzk6fyajaOMyzzbtwdrttHB8mgJI0B9NcAeHdwJtorm7wJuBs4FUM4QoI77zwskV/dvpMRnYGfp9nuHt2+67ZxvGxtPcYkrQAprsCQlXd3bP8vcBH2lmvgNAhg9zScrr7HU/HYWnNh0FuBSdJnbOzKyBMubPRS4Eb2+nLgeOSPDzJwTS3w/zssOKVpOnYAyhJ/dnZFRCOT/J0muHdDcBvAlTVTUkuAb5EcwbxaZ4BrEEM0ss4W/32Ms42pt6eTRwoAAAcC0lEQVReTnsyR8sEUJL6sIsrIFyxi+e8GXjzggUlzbNhJJkaLYeAJUmSOsYewCHwulOSJG3P/42jZQI4hqZ+KWZ7Zpi0mLjzl6TRMQGUJEljzx+N88sEcA48OFaSpPE2l//VXUoaPQlEkiSpY0wAJUmSOsYhYEmSJLp1nKEJoCRJ0hxMlzDu6sod45QwOgQsSZLUMSaAkiRJHWMCKEmS1DFDPwYwyVHAOcBuwPuq6qxhxyBJw+R+TxKM17UJh9oDmGQ34F3A0cCTgeOTPHmYMUjSMLnfkzSOhj0E/ExgfVXdVlXfB9YBxw45BkkaJvd7ksZOqmp4G0teBhxVVb/ezr8CeFZVvbqnzqnAqe3sE4GvDLjZ/YF7BlzHqC2FNsDSaIdtGA+zbcPjq+oxCx3Mrsxmv9eWD7LvWwrv6Uxs49JgGxferPZ7wz4GMNOUbZeBVtVaYO28bTC5tqpWzdf6RmEptAGWRjtsw3hYZG2Ycb8Hg+37FtnrMSe2cWmwjeNj2EPAG4GDeuZXAHcNOQZJGib3e5LGzrATwM8BhyQ5OMkewHHA5UOOQZKGyf2epLEz1CHgqtqW5NXAx2guh3BeVd20wJudt+HkEVoKbYCl0Q7bMB4WTRuGtN9bNK/HAGzj0mAbx8RQTwKRJEnS6HknEEmSpI4xAZQkSeqYJZUAJjkoySeT3JzkpiSvacv3TXJlklvav/uMOtaZJNktyReSfKSdPzjJNW0bLm4PJh9bSZYluTTJl9v34+cX2/uQ5Hfbz9GNSS5K8ojF8D4kOS/J5iQ39pRN+9qn8Y4k65N8MckzRhf5Q3bShr9oP09fTPLhJMt6lp3RtuErSV4wmqiHL8lRbZvXJ1kz6nj60e/+elef1SQntfVvSXLSqNq0M7Pdnyd5eDu/vl2+smcdY/sZ72d/v1jfx37+Hyya97GqlswDOAB4Rjv9k8C/0tx66c+BNW35GuDPRh3rLNryOuBvgY+085cAx7XT7wH+y6hjnCH+84Ffb6f3AJYtpvcBOBD4KrBnz+t/8mJ4H4DnAM8Abuwpm/a1B44B/oHmWnWHA9eMOv5dtOH5wO7t9J/1tOHJwL8ADwcOBm4Fdht1G4bwGu3WtvUJ7XfsX4AnjzquPuLva3+9s88qsC9wW/t3n3Z6n1G3b0pbZ7U/B/4r8J52+jjg4nZ6rD/j/ezvF+P72O//g8XyPo78hV3gN+0y4Jdorqh/QFt2APCVUcc2Q9wrgKuA5wIfab8o9/DQP7+fBz426jh3Ef+j2y9LppQvmveh/cLf0e6Mdm/fhxcslvcBWMn2ydO0rz3wV8Dx09Ub9WNqG6YseylwYTt9BnBGz7KPAT8/6viH8Pps9/mb+jostsdM++udfVaB44G/6infrt6oH/3sz3s/u+1+5562/th+xvvd3y/G97Hf/weL5X1cUkPAvdou158DrgGWV9UmgPbvY0cX2ay8Hfh94Eft/H7A/VW1rZ3fSPOBHFdPAL4B/HU77PG+JI9iEb0PVXUn8BbgdmATsAW4jsX1PvTa2Ws/uWObtFja9CqaXgRYvG0Y1JJp9yz31ztr77i/Dv3sz3/clnb5lrb+OLex3/39onsf5/D/YFG8j0syAUyyF/BB4LVV9a1Rx9OPJC8CNlfVdb3F01Qd5+v37E4zfPfuqvo54AGaIYBFoz1e5ViabvrHAY8Cjp6m6ji/D7Ox2D5bJHkDsA24cLJommpj3YZ5siTa3cf+emftHdvXYQ7780XXRvrf3y+6Ns7h/8GiaOOSSwCTPIxmZ3JhVX2oLb47yQHt8gOAzaOKbxaeDbw4yQZgHc2wwduBZUkmL9w97reS2ghsrKpr2vlLaXYQi+l9eB7w1ar6RlX9APgQ8Assrveh185e+0V1m7L2wPAXASdUO4bCImvDPFr07e5zf72z9o7z69Dv/vzHbWmX7w3cy3i3sd/9/WJ8H/v9f7Ao3scllQAmCXAucHNVvbVn0eXA5BlFJ9EcazKWquqMqlpRVStpDh79RFWdAHwSeFlbbdzb8HXgjiRPbIuOBL7EInofaLr6D0/yyPZzNdmGRfM+TLGz1/5y4MT2zLzDgS2TwzbjJslRwOuBF1fVgz2LLgeOa8+8Oxg4BPjsKGIcskV9i7k57K939ln9GPD8JPu0PTXPb8tGbg778962v6ytX4zxZ3wO+/tF9z7S//+DxfE+jvrgyvl8AP+Bpjv1i8D17eMYmrH3q4Bb2r/7jjrWWbZnNQ+dNfYEmg/KeuDvgIePOr4ZYn86cG37Xvw9zVldi+p9AP4Y+DJwI/ABmjO3xv59AC6iOU7lBzS/OE/Z2WtPMyTxLpqz0W4AVo06/l20YT3N8TOT3+339NR/Q9uGrwBHjzr+Ib5Ox9CcPXsr8IZRx9Nn7H3tr3f1WaU5JnR9+3jlqNu2k/bOuD8HHtHOr2+XP6Hn+WP7Ge9nf79Y38d+/h8slvfRW8FJkiR1zJIaApYkSdLMTAAlSZI6xgRQkiSpY0wAJUmSOsYEUJIkqWNMACVJkjrGBFCSJKljTAAlSZI6xgRQkiSpY0wAJUmSOsYEUJIkqWNMACVJkjrGBFCSJKljTAA1dEk2JHneqOOQJKmrTADVGUl2H3UMkiSNAxNADVWSDwA/BfzvJFuT/H6Sw5P8vyT3J/mXJKt76k8keVOSf07y7SQfT7J/u2x1ko1T1v/j3sUkb0xyaZK/SfIt4OQkP5FkTZJbk3wzySVJ9h3eKyBJ0uiZAGqoquoVwO3AL1fVXsCFwEeBPwX2BX4P+GCSx/Q87deAVwKPBfZo68zWscClwLJ2W78DvAT4ReBxwH3AuwZokiRJi44JoEbtPwNXVNUVVfWjqroSuBY4pqfOX1fVv1bVd4BLgKf3sf7PVNXft+v+DvCbwBuqamNVfQ94I/Ayh4clSV3iPz2N2uOBlyf55Z6yhwGf7Jn/es/0g8Befaz/jmm29+EkP+op+yGwHLizj/VKkrRomQBqFKpn+g7gA1X1G3NYzwPAIydnkuwGPGZKnZoyfwfwqqr65zlsT5KkJcEhYI3C3cAT2um/AX45yQuS7JbkEe3JHStmsZ5/BR6R5IVJHgb8IfDwGZ7zHuDNSR4PkOQxSY6dYzskSVqUTAA1Cv8T+MMk9wO/SnOixh8A36DpoftvzOKzWVVbgP8KvI9m+PYBYOMunwTnAJcDH0/ybeBq4Flza4YkSYtTqqaOkEmSJGkpswdQkiSpY0wAJUmSOsYEUJIkqWNMACVJkjpmrK8DuP/++9fKlSv7es4DDzzAox71qIUJaBHGAeMTi3HsaFxiWapxXHfddfdU1dRrQ0pS5411Arhy5Uquvfbavp4zMTHB6tWrFyagRRgHjE8sxrGjcYllqcaR5GvztjJJWkIcApYkSeoYE0BJkqSOMQGUJEnqGBNASZKkjjEBlCRJ6hgTQEmSpI4Z68vAzMUNd27h5DUfnXX9DWe9cAGjkSRJGj/2AEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdcyMCWCS85JsTnJjT9m+Sa5Mckv7d5+2PEnekWR9ki8meUbPc05q69+S5KSFaY4kSZJmMpsewPcDR00pWwNcVVWHAFe18wBHA4e0j1OBd0OTMAJnAs8CngmcOZk0SpIkabhmTACr6lPAvVOKjwXOb6fPB17SU35BNa4GliU5AHgBcGVV3VtV9wFXsmNSKUmSpCFIVc1cKVkJfKSqntrO319Vy3qW31dV+yT5CHBWVX26Lb8KeD2wGnhEVf1pW/7fge9U1Vum2dapNL2HLF++/LB169b11aDN927h7u/Mvv6hB+7d1/pna+vWrey1114Lsu5+jUssxrGjcYllqcZxxBFHXFdVq+ZthZK0RMz3nUAyTVntonzHwqq1wFqAVatW1erVq/sK4J0XXsbZN8y+WRtO6G/9szUxMUG/sS+UcYnFOHY0LrEYhyR1y1zPAr67Hdql/bu5Ld8IHNRTbwVw1y7KJUmSNGRzTQAvBybP5D0JuKyn/MT2bODDgS1VtQn4GPD8JPu0J388vy2TJEnSkM04VprkIppj+PZPspHmbN6zgEuSnALcDry8rX4FcAywHngQeCVAVd2b5E3A59p6f1JVU08skSRJ0hDMmABW1fE7WXTkNHULOG0n6zkPOK+v6CRJkjTvvBOIJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUscMlAAm+d0kNyW5MclFSR6R5OAk1yS5JcnFSfZo6z68nV/fLl85Hw2QJElSf+acACY5EPgdYFVVPRXYDTgO+DPgbVV1CHAfcEr7lFOA+6rqp4G3tfUkSZI0ZIMOAe8O7Jlkd+CRwCbgucCl7fLzgZe008e287TLj0ySAbcvSZKkPqWq5v7k5DXAm4HvAB8HXgNc3fbykeQg4B+q6qlJbgSOqqqN7bJbgWdV1T1T1nkqcCrA8uXLD1u3bl1fMW2+dwt3f2f29Q89cO++1j9bW7duZa+99lqQdfdrXGIxjh2NSyxLNY4jjjjiuqpaNW8rlKQlYve5PjHJPjS9egcD9wN/Bxw9TdXJDHO63r4dss+qWgusBVi1alWtXr26r7jeeeFlnH3D7Ju14YT+1j9bExMT9Bv7QhmXWIxjR+MSi3FIUrcMMgT8POCrVfWNqvoB8CHgF4Bl7ZAwwArgrnZ6I3AQQLt8b+DeAbYvSZKkORgkAbwdODzJI9tj+Y4EvgR8EnhZW+ck4LJ2+vJ2nnb5J2qQ8WdJkiTNyZwTwKq6huZkjs8DN7TrWgu8HnhdkvXAfsC57VPOBfZry18HrBkgbkmSJM3RnI8BBKiqM4EzpxTfBjxzmrrfBV4+yPYkSZI0OO8EIkmS1DEmgJIkSR0z0BCwJM3VyjUf3aHs9EO3cfI05QAbznrhQockSZ1hD6AkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHXMQAlgkmVJLk3y5SQ3J/n5JPsmuTLJLe3ffdq6SfKOJOuTfDHJM+anCZIkSerHoD2A5wD/WFU/CzwNuBlYA1xVVYcAV7XzAEcDh7SPU4F3D7htSZIkzcGcE8AkjwaeA5wLUFXfr6r7gWOB89tq5wMvaaePBS6oxtXAsiQHzDlySZIkzUmqam5PTJ4OrAW+RNP7dx3wGuDOqlrWU+++qtonyUeAs6rq0235VcDrq+raKes9laaHkOXLlx+2bt26vuLafO8W7v7O7OsfeuDefa1/trZu3cpee+21IOvu17jEYhw7GpdYRhHHDXdu2aFs+Z7s9Ps7l+/qEUcccV1Vrer7iZK0xO0+4HOfAfx2VV2T5BweGu6dTqYp2yH7rKq1NIklq1atqtWrV/cV1DsvvIyzb5h9szac0N/6Z2tiYoJ+Y18o4xKLcexoXGIZRRwnr/noDmWnH7ptp9/fhfquSlIXDXIM4EZgY1Vd085fSpMQ3j05tNv+3dxT/6Ce568A7hpg+5IkSZqDOSeAVfV14I4kT2yLjqQZDr4cOKktOwm4rJ2+HDixPRv4cGBLVW2a6/YlSZI0N4MMAQP8NnBhkj2A24BX0iSVlyQ5BbgdeHlb9wrgGGA98GBbV5IkSUM2UAJYVdcD0x1gfeQ0dQs4bZDtSZIkaXDeCUSSJKljTAAlSZI6ZtBjABe9ldNcimImG8564QJEIkmSNBz2AEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUseYAEqSJHWMCaAkSVLHmABKkiR1jAmgJElSx5gASpIkdYwJoCRJUscMnAAm2S3JF5J8pJ0/OMk1SW5JcnGSPdryh7fz69vlKwfdtiRJkvo3Hz2ArwFu7pn/M+BtVXUIcB9wSlt+CnBfVf008La2niRJkoZsoAQwyQrghcD72vkAzwUubaucD7yknT62naddfmRbX5IkSUOUqpr7k5NLgf8J/CTwe8DJwNVtLx9JDgL+oaqemuRG4Kiq2tguuxV4VlXdM2WdpwKnAixfvvywdevW9RXT5nu3cPd35tykWTn0wL1nrLN161b22msvAG64c8u8r78fvbGMknHsaFxiGUUc030vlu/JTr+/c/leHHHEEddV1aq+nyhJS9zuc31ikhcBm6vquiSrJ4unqVqzWPZQQdVaYC3AqlWravXq1VOr7NI7L7yMs2+Yc7NmZcMJq2esMzExwWTsJ6/56Lyvvx+9sYyScexoXGIZRRzTfS9OP3TbTr+/8/29kKQuGyRTejbw4iTHAI8AHg28HViWZPeq2gasAO5q628EDgI2Jtkd2Bu4d4DtS5IkaQ7mfAxgVZ1RVSuqaiVwHPCJqjoB+CTwsrbaScBl7fTl7Tzt8k/UIOPPkiRJmpOFuA7g64HXJVkP7Aec25afC+zXlr8OWLMA25YkSdIM5uVguaqaACba6duAZ05T57vAy+dje5IkSZq7hT1bYolaOYuTOk4/dFvfJ39Ii9VsvhOSpPHhreAkSZI6xh7Ajuq3x2bDWS9coEgkSdKw2QMoSZLUMSaAkiRJHWMCKEmS1DEmgJIkSR1jAihJktQxngW8ROzqrF6vSShJknrZAyhJktQxJoCSJEkdYwIoSZLUMR4DqAUz9bjEmY5F9G4js7MQ992d+t74XkjS0mYPoCRJUsfYAziGFqKHR42Vaz7a11nRc+kJ6+f9O/3QbazuewuSJA3GHkBJkqSOMQGUJEnqGBNASZKkjjEBlCRJ6hhPAtGi5ckykiTNjT2AkiRJHTPnHsAkBwEXAP8G+BGwtqrOSbIvcDGwEtgA/EpV3ZckwDnAMcCDwMlV9fnBwtdS0tUevXFs9zjGJEmaP4P0AG4DTq+qJwGHA6cleTKwBriqqg4BrmrnAY4GDmkfpwLvHmDbkiRJmqM59wBW1SZgUzv97SQ3AwcCx8KPr217PjABvL4tv6CqCrg6ybIkB7Tr0ZizR0iSpKUjTT424EqSlcCngKcCt1fVsp5l91XVPkk+ApxVVZ9uy68CXl9V105Z16k0PYQsX778sHXr1vUVy+Z7t3D3dwZozDxZvidjEQeMTyyLMY5DD9y77/XfcOeWBYllIS2GOObyXhxxxBHXVdWqAcOSpCVn4LOAk+wFfBB4bVV9qznUb/qq05TtkH1W1VpgLcCqVatq9erVfcXzzgsv4+wbRn9y8+mHbhuLOGB8YlmMcWw4YXXf65/tbeb6jWUhLYY45vJeSJKmN9AeP8nDaJK/C6vqQ23x3ZNDu0kOADa35RuBg3qevgK4a5DtSwvNoW9J0lI055NA2rN6zwVurqq39iy6HDipnT4JuKyn/MQ0Dge2ePyfJEnS8A3SA/hs4BXADUmub8v+ADgLuCTJKcDtwMvbZVfQXAJmPc1lYF45wLYlSZI0R4OcBfxppj+uD+DIaeoXcNpctydJkqT54Z1AJEmSOsYEUJIkqWNMACVJkjrGBFCSJKljTAAlSZI6xgRQkiSpY0wAJUmSOsYEUJIkqWNMACVJkjrGBFCSJKljTAAlSZI6xgRQkiSpY0wAJUmSOsYEUJIkqWNMACVJkjrGBFCSJKljTAAlSZI6xgRQkiSpY0wAJUmSOsYEUJIkqWNMACVJkjpm6AlgkqOSfCXJ+iRrhr19SZKkrhtqAphkN+BdwNHAk4Hjkzx5mDFIkiR13bB7AJ8JrK+q26rq+8A64NghxyBJktRpqarhbSx5GXBUVf16O/8K4FlV9eqeOqcCp7azTwS+0udm9gfumYdwBzUuccD4xGIcOxqXWJZqHI+vqsfM4/okaUnYfcjbyzRl22WgVbUWWDvnDSTXVtWquT5/voxLHDA+sRjHjsYlFuOQpG4Z9hDwRuCgnvkVwF1DjkGSJKnThp0Afg44JMnBSfYAjgMuH3IMkiRJnTbUIeCq2pbk1cDHgN2A86rqpnnezJyHj+fZuMQB4xOLcexoXGIxDknqkKGeBCJJkqTR804gkiRJHWMCKEmS1DFLJgEc5S3mkpyXZHOSG3vK9k1yZZJb2r/7DCGOg5J8MsnNSW5K8ppRxJLkEUk+m+Rf2jj+uC0/OMk1bRwXtycCLbgkuyX5QpKPjDiODUluSHJ9kmvbslF8TpYluTTJl9vPys+PKI4ntq/F5ONbSV47ilgkqWuWRAI4BreYez9w1JSyNcBVVXUIcFU7v9C2AadX1ZOAw4HT2tdh2LF8D3huVT0NeDpwVJLDgT8D3tbGcR9wygLHMek1wM0986OKA+CIqnp6z7XuRvE5OQf4x6r6WeBpNK/N0OOoqq+0r8XTgcOAB4EPjyIWSeqaJZEAMuJbzFXVp4B7pxQfC5zfTp8PvGQIcWyqqs+309+m+cd+4LBjqcbWdvZh7aOA5wKXDisOgCQrgBcC72vnM4o4dmGo702SRwPPAc4FqKrvV9X9w45jGkcCt1bV18YgFkla8pZKAnggcEfP/Ma2bJSWV9UmaBIz4LHD3HiSlcDPAdeMIpZ22PV6YDNwJXArcH9VbWurDOs9ejvw+8CP2vn9RhQHNEnwx5Nc197yEIb/3jwB+Abw1+2w+PuSPGoEcUx1HHBROz3qWCRpyVsqCeCMt5jrkiR7AR8EXltV3xpFDFX1w3ZobwVND+2Tpqu2kDEkeRGwuaqu6y0edhw9nl1Vz6A5VOG0JM8Z0nZ77Q48A3h3Vf0c8AAjHmJtj8F8MfB3o4xDkrpkqSSA43iLubuTHADQ/t08jI0meRhN8ndhVX1olLEAtMOLEzTHJC5LMnnx8WG8R88GXpxkA81hAc+l6REcdhwAVNVd7d/NNMe6PZPhvzcbgY1VdU07fylNQjiyzwhNQvz5qrq7nR9lLJLUCUslARzHW8xdDpzUTp8EXLbQG2yPbzsXuLmq3jqqWJI8JsmydnpP4Hk0xyN+EnjZsOKoqjOqakVVraT5THyiqk4YdhwASR6V5Ccnp4HnAzcy5Pemqr4O3JHkiW3RkcCXhh3HFMfz0PAvI45FkjphydwJJMkxNL07k7eYe/MQt30RsBrYH7gbOBP4e+AS4KeA24H/v307tIkgCKMA/CaQE0jo4AogVwCCArA4DD2gkCS0gjhDERSAwNMDEvUjZpJLTp2AJbn5Prdr9mV3xNvdf26ran+jyG/nuEryluQju5m3x/Q5wMWytNYu04f3T9JfMrZV9dRaW6d/iTtP8p7krqq+/yrHXqbrJA9VdfMfOcY1X8fhaZKXqnpurV1k+XWySd8Us0rymeQ+4zktmWNkOUuf311X1dc4t/g9AZjN0RRAAAAOcyy/gAEAOJACCAAwGQUQAGAyCiAAwGQUQACAySiAAACTUQABACbzA1714wz4j8pfAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 648x432 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.hist(figsize = (9,6), bins = 20)\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And here is the target: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEQCAYAAAC5oaP8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAEIlJREFUeJzt3X+s3XV9x/HnSwriT34WJC2xLHaLmE1kHbC5bE6WUnCzbJEFdaNhTRo3TFg0m7gsoiCJugw344+sGdXCNpGojOrYsEONM5nCZTAUGWmHTpoivdiCYgfy470/zqd6KLe955bbc6Cf5yO5Od/v+/v5fs/729ze1/1+zvecm6pCktSf50y6AUnSZBgAktQpA0CSOmUASFKnDABJ6pQBIEmdMgCk3SR5d5K/n3Qf0v5mAKhbSd6UZCrJQ0nuTfIvSX510n1J47Jg0g1Ik5DkbcBFwFuAG4AfAyuAlcCP5vF5FlTVY/N1PGk+eQWg7iQ5DLgEuKCqPltVP6qqR6vqc1X1p23YIUmuTPLDJHckWTa0fyV52dD6J5K8ty2/JsmWJO9I8j3g40O1tyfZ1q42zh/nOUszMQDUo18GDgWu3cuY1wNXA4cDG4APz+H4LwGOBF4KrBmqHQYsAlYDH0lyxNzaluaXAaAeHQXcP8vUzFer6vqqehy4CnjlHI7/BHBxVT1SVf/Xao8Cl7QrjeuBh4Cf25fmpfliAKhH3weOTrK318C+N7S8Ezh0lvHDpqvq4d2fc7fA2Qm8cMTjSfuFAaAe/QfwMHD2Pu6/E3j+0PpLdtvuR+zqWcEAUHeq6kHgXQzm4c9O8vwkByc5M8kHRjjEbcCbkhyUZAXw6/u1YWk/MQDUpaq6HHgb8BfANHAP8Fbgn0bY/ULgt4EHgDePuI/0jBP/IIwk9ckrAEnqlAEgSZ0yACSpUwaAJHXKAJCkTj2jPw306KOPriVLlky6DUl6Vrnlllvur6qFs417RgfAkiVLmJqamnQbkvSskuR/RxnnFJAkdcoAkKROGQCS1CkDQJI6ZQBIUqcMAEnqlAEgSZ0yACSpU8/oN4I9Wyy56J8n3cIB5Tvve92kW5C64BWAJHXKAJCkThkAktSpkQIgyXeSfCPJbUmmWu3IJBuTbGqPR7R6knwoyeYktyc5eeg4q9r4TUlW7Z9TkiSNYi5XAL9RVSdV1bK2fhFwY1UtBW5s6wBnAkvb1xrgYzAIDOBi4FTgFODiXaEhSRq/pzMFtBJY35bXA2cP1a+sga8Bhyc5DjgD2FhV26tqB7ARWPE0nl+S9DSMGgAFfCHJLUnWtNqxVXUvQHs8ptUXAfcM7bul1fZUf5Ika5JMJZmanp4e/UwkSXMy6vsAXl1VW5McA2xM8t97GZsZarWX+pMLVWuBtQDLli17ynZJ0vwY6Qqgqra2x23AtQzm8O9rUzu0x21t+Bbg+KHdFwNb91KXJE3ArAGQ5AVJXrRrGVgOfBPYAOy6k2cVcF1b3gCc1+4GOg14sE0R3QAsT3JEe/F3eatJkiZglCmgY4Frk+wa/49V9a9JbgauSbIa+C5wTht/PXAWsBnYCZwPUFXbk1wK3NzGXVJV2+ftTCRJczJrAFTV3cArZ6h/Hzh9hnoBF+zhWOuAdXNvU5I033wnsCR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdcoAkKROGQCS1CkDQJI6ZQBIUqcMAEnqlAEgSZ0yACSpUwaAJHXKAJCkThkAktQpA0CSOmUASFKnDABJ6pQBIEmdMgAkqVMGgCR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASeqUASBJnRo5AJIclOTWJJ9v6yck+XqSTUk+leSQVn9uW9/cti8ZOsY7W/2uJGfM98lIkkY3lyuAC4E7h9bfD3ywqpYCO4DVrb4a2FFVLwM+2MaR5ETgXOAVwArgo0kOenrtS5L21UgBkGQx8Drg79p6gNcCn25D1gNnt+WVbZ22/fQ2fiVwdVU9UlXfBjYDp8zHSUiS5m7UK4C/Bv4MeKKtHwU8UFWPtfUtwKK2vAi4B6Btf7CN/0l9hn1+IsmaJFNJpqanp+dwKpKkuZg1AJL8FrCtqm4ZLs8wtGbZtrd9flqoWltVy6pq2cKFC2drT5K0jxaMMObVwOuTnAUcCryYwRXB4UkWtN/yFwNb2/gtwPHAliQLgMOA7UP1XYb3kSSN2axXAFX1zqpaXFVLGLyI+8WqejPwJeANbdgq4Lq2vKGt07Z/saqq1c9tdwmdACwFbpq3M5EkzckoVwB78g7g6iTvBW4Frmj1K4Crkmxm8Jv/uQBVdUeSa4BvAY8BF1TV40/j+SVJT8OcAqCqvgx8uS3fzQx38VTVw8A5e9j/MuCyuTYpSZp/vhNYkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdcoAkKROGQCS1CkDQJI6ZQBIUqcMAEnqlAEgSZ0yACSpUwaAJHXKAJCkThkAktQpA0CSOmUASFKnDABJ6pQBIEmdMgAkqVMGgCR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdcoAkKROzRoASQ5NclOS/0pyR5L3tPoJSb6eZFOSTyU5pNWf29Y3t+1Lho71zla/K8kZ++ukJEmzG+UK4BHgtVX1SuAkYEWS04D3Ax+sqqXADmB1G78a2FFVLwM+2MaR5ETgXOAVwArgo0kOms+TkSSNbtYAqIGH2urB7auA1wKfbvX1wNlteWVbp20/PUla/eqqeqSqvg1sBk6Zl7OQJM3ZSK8BJDkoyW3ANmAj8D/AA1X1WBuyBVjUlhcB9wC07Q8CRw3XZ9hn+LnWJJlKMjU9PT33M5IkjWSkAKiqx6vqJGAxg9/aXz7TsPaYPWzbU33351pbVcuqatnChQtHaU+StA/mdBdQVT0AfBk4DTg8yYK2aTGwtS1vAY4HaNsPA7YP12fYR5I0ZqPcBbQwyeFt+XnAbwJ3Al8C3tCGrQKua8sb2jpt+xerqlr93HaX0AnAUuCm+ToRSdLcLJh9CMcB69sdO88Brqmqzyf5FnB1kvcCtwJXtPFXAFcl2czgN/9zAarqjiTXAN8CHgMuqKrH5/d0JEmjmjUAqup24FUz1O9mhrt4quph4Jw9HOsy4LK5tylJmm++E1iSOmUASFKnDABJ6pQBIEmdMgAkqVMGgCR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdcoAkKROGQCS1CkDQJI6ZQBIUqcMAEnqlAEgSZ0yACSpUwaAJHXKAJCkThkAktQpA0CSOmUASFKnDABJ6pQBIEmdMgAkqVMGgCR1ygCQpE7NGgBJjk/ypSR3JrkjyYWtfmSSjUk2tccjWj1JPpRkc5Lbk5w8dKxVbfymJKv232lJkmYzyhXAY8Dbq+rlwGnABUlOBC4CbqyqpcCNbR3gTGBp+1oDfAwGgQFcDJwKnAJcvCs0JEnjN2sAVNW9VfWfbfmHwJ3AImAlsL4NWw+c3ZZXAlfWwNeAw5McB5wBbKyq7VW1A9gIrJjXs5EkjWxOrwEkWQK8Cvg6cGxV3QuDkACOacMWAfcM7bal1fZU3/051iSZSjI1PT09l/YkSXMwcgAkeSHwGeBPquoHexs6Q632Un9yoWptVS2rqmULFy4ctT1J0hyNFABJDmbww/8fquqzrXxfm9qhPW5r9S3A8UO7Lwa27qUuSZqAUe4CCnAFcGdVXT60aQOw606eVcB1Q/Xz2t1ApwEPtimiG4DlSY5oL/4ubzVJ0gQsGGHMq4E/AL6R5LZW+3PgfcA1SVYD3wXOaduuB84CNgM7gfMBqmp7kkuBm9u4S6pq+7ychSRpzmYNgKr6KjPP3wOcPsP4Ai7Yw7HWAevm0qAkaf8Y5QpA0rPZuw+bdAcHjnc/OOkO5pUfBSFJnTIAJKlTBoAkdcoAkKROGQCS1CkDQJI6ZQBIUqcMAEnqlAEgSZ0yACSpUwaAJHXKAJCkThkAktQpA0CSOmUASFKnDABJ6pQBIEmdMgAkqVMGgCR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdcoAkKROGQCS1CkDQJI6NWsAJFmXZFuSbw7VjkyyMcmm9nhEqyfJh5JsTnJ7kpOH9lnVxm9Ksmr/nI4kaVSjXAF8AlixW+0i4MaqWgrc2NYBzgSWtq81wMdgEBjAxcCpwCnAxbtCQ5I0GbMGQFV9Bdi+W3klsL4trwfOHqpfWQNfAw5PchxwBrCxqrZX1Q5gI08NFUnSGO3rawDHVtW9AO3xmFZfBNwzNG5Lq+2p/hRJ1iSZSjI1PT29j+1JkmYz3y8CZ4Za7aX+1GLV2qpaVlXLFi5cOK/NSZJ+al8D4L42tUN73NbqW4Djh8YtBrbupS5JmpB9DYANwK47eVYB1w3Vz2t3A50GPNimiG4Alic5or34u7zVJEkTsmC2AUk+CbwGODrJFgZ387wPuCbJauC7wDlt+PXAWcBmYCdwPkBVbU9yKXBzG3dJVe3+wrIkaYxmDYCqeuMeNp0+w9gCLtjDcdYB6+bUnSRpv/GdwJLUKQNAkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdcoAkKROGQCS1CkDQJI6ZQBIUqcMAEnqlAEgSZ0yACSpUwaAJHXKAJCkThkAktQpA0CSOmUASFKnDABJ6pQBIEmdMgAkqVMGgCR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASeqUASBJnTIAJKlTBoAkdWrsAZBkRZK7kmxOctG4n1+SNDDWAEhyEPAR4EzgROCNSU4cZw+SpIFxXwGcAmyuqrur6sfA1cDKMfcgSQIWjPn5FgH3DK1vAU4dHpBkDbCmrT6U5K4x9daDo4H7J93EbPL+SXegCXhWfG/ynky6g1G9dJRB4w6Amf716kkrVWuBteNppy9Jpqpq2aT7kHbn9+ZkjHsKaAtw/ND6YmDrmHuQJDH+ALgZWJrkhCSHAOcCG8bcgySJMU8BVdVjSd4K3AAcBKyrqjvG2UPnnFrTM5XfmxOQqpp9lCTpgOM7gSWpUwaAJHXKAJCkThkAksYuyVuTvLgt/22Sm5KcPum+emMAHOCSLE5ybZLpJPcl+UySxZPuS91bU1U/SLKcwScE/BHwgQn31B0D4MD3cQbvtTiOwX+0z7WaNEm7bj88E/h4Vd2CP4/GzttAD3BJbquqk2arSeOU5EoGn//zs8AvMPjh/5WqOnmijXVm3J8FpPG7P8nvA59s628Evj/BfiSA84FfZPDpwDuTHA2snnBP3fGS68D3h8DvAd8D7gXe0GrSxFTV48DPMJj7B3ge/jwaO6eAJI1dkg8DBwO/VlUvT3IkcENV/dKEW+uKU0AHqCTv2svmqqpLx9aM9FS/UlUnJ7kVoKq2tw+I1BgZAAeuH81QewGDedajAANAk/RokufQ7gZKchTwxGRb6o9TQB1I8iLgQgY//K8B/qqqtk22K/UsyXnA7wDLgHUMXqd6T1VdPdHGOmMAHMDavOrbgDcD64G/qaodk+1KPUtyPfDHVfWdJK8AfpPBXwr8t6r65mS7649TQAeoJH8J/C6Dz1n/+ap6aMItSQCfAL6QZD3wAf8eyGR5BXCASvIE8AjwGE/+u8th8CLwiyfSmLqX5AXAu4AVwFUMzf1X1eWT6qtHXgEcoKrKe6r1TPUog5sUngu8CF/8nRgDQNLYJFkBXM7g86lOrqqdE26pa04BSRqbJP8OvMW5/2cGA0CSOuU8sSR1ygCQpE4ZAJLUKQNAkjplAEhSpwwASerU/wOQuzrxQtz1+gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['Churn'].value_counts().plot(kind='bar')\n",
    "plt.title('Churn')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train-test split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop('customerID', axis=1, inplace=True)\n",
    "X = df.drop('Churn', axis=1)\n",
    "y = df['Churn']\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imputing missing values and training some models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import NearestCentroid\n",
    "from sklearn.svm import LinearSVC\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.preprocessing import StandardScaler, OneHotEncoder\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.model_selection import cross_val_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score for Logistic Regression: 0.7962957476755843\n",
      "Score for Nearest Centroid: 0.5187528865994284\n",
      "Score for Linear SVC: 0.7001111860490169\n"
     ]
    }
   ],
   "source": [
    "si = SimpleImputer(strategy=\"mean\")\n",
    "ohe = OneHotEncoder(categories=\"auto\", handle_unknown = 'ignore')\n",
    "logr = LogisticRegression()\n",
    "nc = NearestCentroid()\n",
    "lsvc = LinearSVC()\n",
    "\n",
    "Xt_cat = X_train.dtypes == object\n",
    "pipe_prep_cont = Pipeline([('imputer', si)])\n",
    "pipe_prep_cat = Pipeline([('ohe', ohe)])\n",
    "\n",
    "prep = ColumnTransformer([('continuous', pipe_prep_cont, ~Xt_cat),('categorical', pipe_prep_cat, Xt_cat)])\n",
    "\n",
    "pipe_logr = Pipeline([('preprocessing', prep), ('logr', logr)])\n",
    "pipe_nc = Pipeline([('preprocessing', prep), ('nc', nc)])\n",
    "pipe_lsvc = Pipeline([('preprocessing', prep), ('lsvc', lsvc)])\n",
    "\n",
    "s_logr = np.mean(cross_val_score(pipe_logr, X_train, y_train, cv=10))\n",
    "print(\"Score for Logistic Regression: \" + str(s_logr))\n",
    "\n",
    "s_nc = np.mean(cross_val_score(pipe_nc, X_train, y_train, cv=10))\n",
    "print(\"Score for Nearest Centroid: \" + str(s_nc))\n",
    "\n",
    "s_lsvc = np.mean(cross_val_score(pipe_lsvc, X_train, y_train, cv=10))\n",
    "print(\"Score for Linear SVC: \" + str(s_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The results vary a lot between the different classifiers. Logistic regression works the best, then Linear SVC, and Nearest Centroid performs very poorly. Let's now try with scaling the data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score for Logistic Regression with standard scaler: 0.7970536814557299\n",
      "Score for Nearest Centroid with standard scaler: 0.7334444922079615\n",
      "Score for Linear SVC with standard scaler: 0.7976225806777706\n"
     ]
    }
   ],
   "source": [
    "ssc = StandardScaler()\n",
    "pipe_prep_cont = Pipeline([('imputer', si), ('scaler', ssc)])\n",
    "prep = ColumnTransformer([('continuous', pipe_prep_cont, ~Xt_cat),('categorical', pipe_prep_cat, Xt_cat)])\n",
    "\n",
    "pipe_logr = Pipeline([('preprocessing', prep), ('logr', logr)])\n",
    "pipe_nc = Pipeline([('preprocessing', prep), ('nc', nc)])\n",
    "pipe_lsvc = Pipeline([('preprocessing', prep), ('lsvc', lsvc)])\n",
    "\n",
    "s_logr = np.mean(cross_val_score(pipe_logr, X_train, y_train, cv=10))\n",
    "print(\"Score for Logistic Regression with standard scaler: \" + str(s_logr))\n",
    "\n",
    "s_nc = np.mean(cross_val_score(pipe_nc, X_train, y_train, cv=10))\n",
    "print(\"Score for Nearest Centroid with standard scaler: \" + str(s_nc))\n",
    "\n",
    "s_lsvc = np.mean(cross_val_score(pipe_lsvc, X_train, y_train, cv=10))\n",
    "print(\"Score for Linear SVC with standard scaler: \" + str(s_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scaling the data actually significantly improves both Nearest Centroid and SVM. With scaling, Nearest Centroid still performs the worst but linear SVC and Logistic Regression are roughly equivalent. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Tuning the parameters using a grid search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Score for Logistic Regression with Standard Scaling and Grid Search CV: 0.7989397955319955\n",
      "Score for Nearest Centroids with Standard Scaling and Grid Search CV: 0.733434305187429\n",
      "Score for Linear SVC with Standard Scaling and Grid Search CV: 0.798750473305566\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "param_grid_logr = {'logr__C': np.logspace(-3, 2, 6)}\n",
    "param_grid_nc = {'nc__shrink_threshold': np.linspace(0, 2, 20)}\n",
    "param_grid_lsvc = {'lsvc__C': np.logspace(-3, 2, 6)}\n",
    "\n",
    "grid_logr = GridSearchCV(pipe_logr, param_grid_logr, cv=10)\n",
    "_ = grid_logr.fit(X_train, y_train)\n",
    "\n",
    "grid_nc = GridSearchCV(pipe_nc, param_grid_nc, cv=10)\n",
    "_ = grid_nc.fit(X_train, y_train)\n",
    "\n",
    "grid_lsvc = GridSearchCV(pipe_lsvc, param_grid_lsvc, cv=10)\n",
    "_ = grid_lsvc.fit(X_train, y_train)\n",
    "\n",
    "s_logr = grid_logr.best_score_\n",
    "print(\"Score for Logistic Regression with Standard Scaling and Grid Search CV: \" + str(s_logr))\n",
    "\n",
    "s_nc = grid_nc.best_score_\n",
    "print(\"Score for Nearest Centroids with Standard Scaling and Grid Search CV: \" + str(s_nc))\n",
    "\n",
    "s_lsvc = grid_lsvc.best_score_\n",
    "print(\"Score for Linear SVC with Standard Scaling and Grid Search CV: \" + str(s_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The results barely improve with GridSearch. It looks like the default parameters give us good enough models. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now visualize the training and test scores depending on the parameters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results_logr = pd.DataFrame(grid_logr.cv_results_)\n",
    "results_logr.plot('param_logr__C', 'mean_train_score')\n",
    "results_logr.plot('param_logr__C', 'mean_test_score', ax=plt.gca())\n",
    "plt.legend()\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.title(\"Logistic Regression\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results_nc = pd.DataFrame(grid_nc.cv_results_)\n",
    "results_nc.plot('param_nc__shrink_threshold', 'mean_train_score')\n",
    "results_nc.plot('param_nc__shrink_threshold', 'mean_test_score', ax=plt.gca())\n",
    "plt.legend()\n",
    "plt.xlabel(\"Shrink Threshold\")\n",
    "plt.title(\"Nearest Centroid\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results_lsvc = pd.DataFrame(grid_lsvc.cv_results_)\n",
    "results_lsvc.plot('param_lsvc__C', 'mean_train_score')\n",
    "results_lsvc.plot('param_lsvc__C', 'mean_test_score', ax=plt.gca())\n",
    "plt.legend()\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.title(\"Linear SVC\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using a k-fold cross-validation with shuffling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's first view the parameters currently found by GridSearchCV: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for Logistic Regression: {'logr__C': 0.01}\n",
      "Best parameters for Nearest Centroids: {'nc__shrink_threshold': 0.0}\n",
      "Best parameters for Linear SVC: {'lsvc__C': 0.1}\n"
     ]
    }
   ],
   "source": [
    "bp_logr = grid_logr.best_params_\n",
    "print(\"Best parameters for Logistic Regression: \" + str(bp_logr))\n",
    "\n",
    "bp_nc = grid_nc.best_params_\n",
    "print(\"Best parameters for Nearest Centroids: \" + str(bp_nc))\n",
    "\n",
    "bp_lsvc = grid_lsvc.best_params_\n",
    "print(\"Best parameters for Linear SVC: \" + str(bp_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now find them again using a k-fold with shuffling cross-validation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for Logistic Regression with k-fold shuffling CV: {'logr__C': 0.01}\n",
      "Best parameters for Nearest Centroids with k-fold shuffling CV: {'nc__shrink_threshold': 0.0}\n",
      "Best parameters for Linear SVC with k-fold shuffling CV: {'lsvc__C': 0.01}\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import KFold\n",
    "\n",
    "cvs=KFold(n_splits=10, shuffle=True)\n",
    "\n",
    "grid_logr = GridSearchCV(pipe_logr, param_grid_logr, cv=cvs)\n",
    "_ = grid_logr.fit(X_train, y_train)\n",
    "\n",
    "grid_nc = GridSearchCV(pipe_nc, param_grid_nc, cv=cvs)\n",
    "_ = grid_nc.fit(X_train, y_train)\n",
    "\n",
    "grid_lsvc = GridSearchCV(pipe_lsvc, param_grid_lsvc, cv=cvs)\n",
    "_ = grid_lsvc.fit(X_train, y_train)\n",
    "\n",
    "bp_logr = grid_logr.best_params_\n",
    "print(\"Best parameters for Logistic Regression with k-fold shuffling CV: \" + str(bp_logr))\n",
    "\n",
    "bp_nc = grid_nc.best_params_\n",
    "print(\"Best parameters for Nearest Centroids with k-fold shuffling CV: \" + str(bp_nc))\n",
    "\n",
    "bp_lsvc = grid_lsvc.best_params_\n",
    "print(\"Best parameters for Linear SVC with k-fold shuffling CV: \" + str(bp_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the parameters don't change, except for Linear SVC where the best C is now 0.001 instead of 1. However, let's view the corresponding plot:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results_lsvc = pd.DataFrame(grid_lsvc.cv_results_)\n",
    "results_lsvc.plot('param_lsvc__C', 'mean_train_score')\n",
    "results_lsvc.plot('param_lsvc__C', 'mean_test_score', ax=plt.gca())\n",
    "plt.legend()\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.title(\"Linear SVC with K-Fold and shuffling\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the difference between 0.001 and 1 is very minimal in terms of performance - as it was in the previous plot. So the change in best parameter isn't really significant."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now add a random state to our K-Fold."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for Logistic Regression with k-fold shuffling CV and random state on CV: {'logr__C': 100.0}\n",
      "Best parameters for Nearest Centroids with k-fold shuffling CV and random state on CV: {'nc__shrink_threshold': 0.0}\n",
      "Best parameters for Linear SVC with k-fold shuffling CV and random state on CV: {'lsvc__C': 0.1}\n"
     ]
    }
   ],
   "source": [
    "cvs=KFold(n_splits=10, shuffle=True, random_state=1)\n",
    "\n",
    "grid_logr = GridSearchCV(pipe_logr, param_grid_logr, cv=cvs)\n",
    "_ = grid_logr.fit(X_train, y_train)\n",
    "\n",
    "grid_nc = GridSearchCV(pipe_nc, param_grid_nc, cv=cvs)\n",
    "_ = grid_nc.fit(X_train, y_train)\n",
    "\n",
    "grid_lsvc = GridSearchCV(pipe_lsvc, param_grid_lsvc, cv=cvs)\n",
    "_ = grid_lsvc.fit(X_train, y_train)\n",
    "\n",
    "bp_logr = grid_logr.best_params_\n",
    "print(\"Best parameters for Logistic Regression with k-fold shuffling CV and random state on CV: \" \n",
    "      + str(bp_logr))\n",
    "\n",
    "bp_nc = grid_nc.best_params_\n",
    "print(\"Best parameters for Nearest Centroids with k-fold shuffling CV and random state on CV: \" \n",
    "      + str(bp_nc))\n",
    "\n",
    "bp_lsvc = grid_lsvc.best_params_\n",
    "print(\"Best parameters for Linear SVC with k-fold shuffling CV and random state on CV: \" \n",
    "      + str(bp_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The scores don't change at all from last time - which is reassuring since it means our models are not heavily affected by randomness."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, let's also add a random state for the splitting of the data between train set and test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for Logistic Regression with k-fold shuffling CV and random state on CV and train/test split:\n",
      " {'logr__C': 0.01}\n",
      "Best parameters for Nearest Centroids with k-fold shuffling CV and random state on CV and train/test split:\n",
      " {'nc__shrink_threshold': 0.0}\n",
      "Best parameters for Linear SVC with k-fold shuffling CV and random state on CV and train/test split:\n",
      " {'lsvc__C': 0.001}\n"
     ]
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=1)\n",
    "grid_logr = GridSearchCV(pipe_logr, param_grid_logr, cv=cvs)\n",
    "_ = grid_logr.fit(X_train, y_train)\n",
    "\n",
    "grid_nc = GridSearchCV(pipe_nc, param_grid_nc, cv=cvs)\n",
    "_ = grid_nc.fit(X_train, y_train)\n",
    "\n",
    "grid_lsvc = GridSearchCV(pipe_lsvc, param_grid_lsvc, cv=cvs)\n",
    "_ = grid_lsvc.fit(X_train, y_train)\n",
    "\n",
    "bp_logr = grid_logr.best_params_\n",
    "print(\"Best parameters for Logistic Regression with k-fold shuffling CV and random state on CV and train/test split:\\n \" \n",
    "      + str(bp_logr))\n",
    "\n",
    "bp_nc = grid_nc.best_params_\n",
    "print(\"Best parameters for Nearest Centroids with k-fold shuffling CV and random state on CV and train/test split:\\n \" \n",
    "      + str(bp_nc))\n",
    "\n",
    "bp_lsvc = grid_lsvc.best_params_\n",
    "print(\"Best parameters for Linear SVC with k-fold shuffling CV and random state on CV and train/test split:\\n \" \n",
    "      + str(bp_lsvc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This time, the best C parameter for Logistic Regression changes from 1 to 0.01. But again, let's plot the evolution of the scores according to that parameter:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results_logr = pd.DataFrame(grid_logr.cv_results_)\n",
    "results_logr.plot('param_logr__C', 'mean_train_score')\n",
    "results_logr.plot('param_logr__C', 'mean_test_score', ax=plt.gca())\n",
    "plt.legend()\n",
    "plt.xscale(\"log\")\n",
    "plt.xlabel(\"C\")\n",
    "plt.title(\"Logistic Regression with K-Fold and shuffling\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, we can see that the curve is rougly constant from 0.01 on, so it's not really a surprise that a slight change (e.g. changing the train-test split) would affect the maximum of the curve, since it's always going to be on that straight line between 0.01 and 100."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing the coefficients for Logistic Regression and SVC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We display a 'dot plot' of feature coefficients for both models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "coeffs_logr = grid_logr.best_estimator_.named_steps['logr'].coef_\n",
    "coeffs_lsvc = grid_lsvc.best_estimator_.named_steps['lsvc'].coef_\n",
    "plt.plot(coeffs_logr[0], 'o', label='Logistic Regression')\n",
    "plt.plot(coeffs_lsvc[0], 'o', label='Linear SVC')\n",
    "plt.legend()\n",
    "ax = plt.gca()\n",
    "ax.set_xticks(range(len(coeffs_logr[0])), minor=True)\n",
    "plt.gca().xaxis.grid(True, which='minor')\n",
    "plt.title(\"Distribution of feature coefficients for LogReg and SVC\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can wee that the general distribution of the feature coefficients is roughly similar between the two models (in particular, features 1, 17, 19, 37 and 44 are important in both), even though the amplitude of the coefficients is generally larger in Logistic Regression."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}