Bootstrap-t-student/bootstrap-t.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "Bootstrapowa wersja testu t.\n",
    "Implementacja powinna obejmować test dla jednej próby, dla dwóch prób niezależnych oraz dla dwóch prób zależnych.\n",
    "W każdej sytuacji oczekiwanym wejście jest zbiór danych w odpowiednim formacie, a wyjściem p-wartość oraz ostateczna decyzja.\n",
    "Dodatkowo powinien być rysowany odpowiedni rozkład statystyki testowej."
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "Zbiór danych          - ???\n",
    "Hipoteza zerowa       - ???\n",
    "Hipoteza alternatywna - ???\n",
    "\n",
    "Dla każdego z 3 testów inne\n",
    "https://www.jmp.com/en_ch/statistics-knowledge-portal/t-test.html"
   ],
   "metadata": {
    "collapsed": false
   }
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from math import sqrt\n",
    "from scipy.stats import sem\n",
    "from scipy.stats import t\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "def generate_bootstraps(data, n_bootstraps=100):\n",
    "    data_size = data.shape[0]\n",
    "    for _ in range(n_bootstraps):\n",
    "        indices =  np.random.choice(len(data), size=data_size)\n",
    "        yield data.iloc[indices, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "outputs": [],
   "source": [
    "def get_t_stat(data1, data2):\n",
    "    mean1 = np.mean(data1)\n",
    "    mean2 = np.mean(data2)\n",
    "    sem1 = sem(data1)\n",
    "    sem2 = sem(data2)\n",
    "\n",
    "    sed = sqrt(sem1**2.0 + sem2**2.0)\n",
    "    # To jest wzór chyba tylko dla jednego przypadku\n",
    "    return (mean1 - mean2) / sed"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "def independent_t_test(data, columns, alpha=0.05):\n",
    "    t_stat_sum = 0\n",
    "    for sample in generate_bootstraps(data):\n",
    "        t_stat_sum += get_t_stat(sample[columns[0]], sample[columns[1]])\n",
    "\n",
    "    data_size = data.shape[0]\n",
    "    t_stat = t_stat_sum / data_size\n",
    "    df = 2 * data_size - 2\n",
    "    cv = t.ppf(1.0 - alpha, df)\n",
    "    p = (1.0 - t.cdf(abs(t_stat), df)) * 2.0\n",
    "    return t_stat, df, cv, p"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "outputs": [],
   "source": [
    "def make_decision(data, columns, alpha=0.05):\n",
    "    t_stat, df, cv, p = independent_t_test(data, columns, alpha)\n",
    "    print(f't: {t_stat}, df: {df}, cv: {cv}, p: {p}\\n')\n",
    "    if abs(t_stat) <= cv:\n",
    "\t    print('Accept null hypothesis that the means are equal.')\n",
    "    else:\n",
    "        print('Reject the null hypothesis that the means are equal.')\n",
    "    if p > alpha:\n",
    "        print('Accept null hypothesis that the means are equal.')\n",
    "    else:\n",
    "\t    print('Reject the null hypothesis that the means are equal.')"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "outputs": [
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def draw_distribution():\n",
    "    \"\"\"Funkcja rysuje rozkład statystyki testowej\"\"\"\n",
    "    dummy = np.random.normal(170, 10, 500)\n",
    "    plt.hist(dummy)\n",
    "    plt.show()\n",
    "    pass\n",
    "draw_distribution()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "t: 6.893215520199072, df: 998, cv: 1.6463818766348755, p: 9.657386002004387e-12\n",
      "\n",
      "Reject the null hypothesis that the means are equal.\n",
      "Reject the null hypothesis that the means are equal.\n"
     ]
    }
   ],
   "source": [
    "dataset = pd.read_csv('experiment_data.csv')\n",
    "make_decision(dataset, ['Weight', 'Age'])"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "11938c6bc6919ae2720b4d5011047913343b08a43b18698fd82dedb0d4417594"
  },
  "kernelspec": {
   "display_name": "Python 3.9.1 64-bit",
   "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": "3.9.1"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}