Ch05-ex02.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## インポート"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "\n",
    "np.random.seed(1234)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## データの読み込み"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>A</th>\n",
       "      <th>Score</th>\n",
       "      <th>Y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>69</td>\n",
       "      <td>0.286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>145</td>\n",
       "      <td>0.196</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>125</td>\n",
       "      <td>0.261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>86</td>\n",
       "      <td>0.109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>158</td>\n",
       "      <td>0.230</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   A  Score      Y\n",
       "0  0     69  0.286\n",
       "1  1    145  0.196\n",
       "2  0    125  0.261\n",
       "3  1     86  0.109\n",
       "4  1    158  0.230"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('chap05/input/data-attendance-1.txt')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## モデル式の記述\n",
    "\n",
    "$\\mu[n] = \\beta_1 + \\beta_2 A[n] + \\beta_3 Score[n] $\n",
    "\n",
    "$Y[n]$ ~ $ Normal( \\mu[n], \\sigma )$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Stanでの実装"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_d0d7a2793aebf17a8db26573daf4d5ed NOW.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Inference for Stan model: anon_model_d0d7a2793aebf17a8db26573daf4d5ed.\n",
       "4 chains, each with iter=2000; warmup=1000; thin=1; \n",
       "post-warmup draws per chain=1000, total post-warmup draws=4000.\n",
       "\n",
       "              mean se_mean     sd   2.5%    25%    50%    75%  97.5%  n_eff   Rhat\n",
       "b1            0.12  8.3e-4   0.03   0.05    0.1   0.12   0.15   0.19   1718    1.0\n",
       "b2           -0.14  3.0e-4   0.01  -0.17  -0.15  -0.14  -0.13  -0.11   2538    1.0\n",
       "b3            0.32  1.3e-3   0.05   0.22   0.29   0.32   0.36   0.43   1730    1.0\n",
       "sigma         0.05  1.1e-4 5.6e-3   0.04   0.05   0.05   0.06   0.06   2385    1.0\n",
       "mu[0]         0.24  4.0e-4   0.02    0.2   0.22   0.24   0.25   0.27   1865    1.0\n",
       "mu[1]         0.22  2.6e-4   0.01   0.19   0.21   0.22   0.22   0.24   2579    1.0\n",
       "mu[2]         0.33  1.7e-4 9.7e-3   0.31   0.32   0.33   0.33   0.35   3287    1.0\n",
       "mu[3]         0.12  3.2e-4   0.02   0.09   0.11   0.12   0.13   0.15   2344    1.0\n",
       "mu[4]         0.24  3.2e-4   0.02   0.21   0.23   0.24   0.25   0.27   2289    1.0\n",
       "mu[5]         0.34  1.8e-4   0.01   0.32   0.33   0.34   0.35   0.36   3258    1.0\n",
       "mu[6]          0.3  1.9e-4   0.01   0.28    0.3    0.3   0.31   0.32   2858    1.0\n",
       "mu[7]         0.22  2.7e-4   0.01   0.19   0.21   0.22   0.23   0.24   2527    1.0\n",
       "mu[8]         0.36  2.2e-4   0.01   0.34   0.35   0.36   0.37   0.38   2916    1.0\n",
       "mu[9]         0.36  2.1e-4   0.01   0.34   0.35   0.36   0.37   0.38   2945    1.0\n",
       "mu[10]        0.21  2.5e-4   0.01   0.18    0.2   0.21   0.22   0.23   2687    1.0\n",
       "mu[11]        0.35  1.9e-4   0.01   0.32   0.34   0.35   0.35   0.37   3171    1.0\n",
       "mu[12]        0.17  2.1e-4   0.01   0.15   0.16   0.17   0.18   0.19   3014    1.0\n",
       "mu[13]         0.3  1.9e-4   0.01   0.28    0.3    0.3   0.31   0.32   2858    1.0\n",
       "mu[14]         0.3  2.0e-4   0.01   0.28   0.29    0.3   0.31   0.32   2748    1.0\n",
       "mu[15]        0.14  2.6e-4   0.01   0.11   0.13   0.14   0.15   0.17   2620    1.0\n",
       "mu[16]        0.31  1.8e-4 9.9e-3   0.29    0.3   0.31   0.32   0.33   2977    1.0\n",
       "mu[17]        0.26  3.3e-4   0.01   0.23   0.25   0.26   0.27   0.29   1988    1.0\n",
       "mu[18]        0.42  4.3e-4   0.02   0.38   0.41   0.42   0.44   0.46   2120    1.0\n",
       "mu[19]        0.23  3.1e-4   0.01    0.2   0.22   0.23   0.24   0.26   2326    1.0\n",
       "mu[20]        0.12  3.2e-4   0.02   0.09   0.11   0.12   0.13   0.15   2344    1.0\n",
       "mu[21]        0.16  2.3e-4   0.01   0.13   0.15   0.16   0.16   0.18   2842    1.0\n",
       "mu[22]        0.15  2.4e-4   0.01   0.13   0.14   0.15   0.16   0.18   2794    1.0\n",
       "mu[23]        0.21  2.5e-4   0.01   0.19    0.2   0.21   0.22   0.24   2632    1.0\n",
       "mu[24]        0.17  2.2e-4   0.01   0.15   0.16   0.17   0.18   0.19   3004    1.0\n",
       "mu[25]        0.19  2.2e-4   0.01   0.16   0.18   0.19    0.2   0.21   2981    1.0\n",
       "mu[26]        0.32  1.7e-4 9.7e-3    0.3   0.31   0.32   0.32   0.34   3156    1.0\n",
       "mu[27]        0.32  1.7e-4 9.7e-3    0.3   0.31   0.32   0.32   0.34   3156    1.0\n",
       "mu[28]        0.39  2.9e-4   0.01   0.36   0.38   0.39   0.39   0.41   2457    1.0\n",
       "mu[29]        0.31  1.8e-410.0e-3   0.29    0.3   0.31   0.31   0.33   2937    1.0\n",
       "mu[30]        0.25  3.5e-4   0.02   0.22   0.24   0.25   0.26   0.28   2177    1.0\n",
       "mu[31]         0.1  3.8e-4   0.02   0.07   0.09    0.1   0.11   0.13   2170    1.0\n",
       "mu[32]         0.2  2.4e-4   0.01   0.18    0.2    0.2   0.21   0.23   2768    1.0\n",
       "mu[33]        0.18  2.1e-4   0.01   0.16   0.17   0.18   0.19    0.2   3022    1.0\n",
       "mu[34]        0.33  1.7e-4 9.7e-3   0.31   0.32   0.33   0.33   0.35   3297    1.0\n",
       "mu[35]        0.34  1.7e-410.0e-3   0.32   0.33   0.34   0.34   0.36   3286    1.0\n",
       "mu[36]        0.15  2.4e-4   0.01   0.13   0.14   0.15   0.16   0.18   2769    1.0\n",
       "mu[37]         0.3  1.9e-4   0.01   0.28    0.3    0.3   0.31   0.32   2819    1.0\n",
       "mu[38]        0.27  2.9e-4   0.01   0.24   0.26   0.27   0.28   0.29   2099    1.0\n",
       "mu[39]        0.27  2.9e-4   0.01   0.24   0.26   0.27   0.28   0.29   2081    1.0\n",
       "mu[40]        0.33  1.7e-4 9.9e-3   0.31   0.33   0.33   0.34   0.35   3301    1.0\n",
       "mu[41]        0.34  1.8e-4   0.01   0.32   0.33   0.34   0.35   0.36   3258    1.0\n",
       "mu[42]        0.32  1.7e-4 9.7e-3    0.3   0.32   0.32   0.33   0.34   3237    1.0\n",
       "mu[43]        0.36  2.2e-4   0.01   0.34   0.36   0.36   0.37   0.39   2859    1.0\n",
       "mu[44]        0.42  4.1e-4   0.02   0.38   0.41   0.42   0.43   0.46   2147    1.0\n",
       "mu[45]        0.29  2.1e-4   0.01   0.27   0.29   0.29    0.3   0.32   2570    1.0\n",
       "mu[46]        0.21  5.1e-4   0.02   0.16   0.19   0.21   0.22   0.25   1783    1.0\n",
       "mu[47]        0.37  2.4e-4   0.01   0.34   0.36   0.37   0.38   0.39   2753    1.0\n",
       "mu[48]        0.28  2.4e-4   0.01   0.26   0.28   0.28   0.29   0.31   2349    1.0\n",
       "mu[49]        0.14  2.6e-4   0.01   0.11   0.13   0.14   0.15   0.17   2620    1.0\n",
       "y_pred[0]     0.24  9.4e-4   0.06   0.13    0.2   0.24   0.27   0.34   3497    1.0\n",
       "y_pred[1]     0.21  8.6e-4   0.05   0.11   0.18   0.21   0.25   0.32   4000    1.0\n",
       "y_pred[2]     0.33  8.3e-4   0.05   0.22   0.29   0.33   0.36   0.43   4000    1.0\n",
       "y_pred[3]     0.12  8.8e-4   0.05 9.4e-3   0.08   0.12   0.16   0.22   3786    1.0\n",
       "y_pred[4]     0.24  9.2e-4   0.05   0.13    0.2   0.24   0.27   0.35   3528    1.0\n",
       "y_pred[5]     0.34  9.0e-4   0.05   0.24    0.3   0.34   0.37   0.44   3395    1.0\n",
       "y_pred[6]      0.3  8.6e-4   0.05    0.2   0.27    0.3   0.34   0.41   3795    1.0\n",
       "y_pred[7]     0.22  8.8e-4   0.05   0.11   0.18   0.22   0.25   0.32   3716    1.0\n",
       "y_pred[8]     0.36  8.6e-4   0.05   0.26   0.32   0.36    0.4   0.46   3747    1.0\n",
       "y_pred[9]     0.36  9.0e-4   0.05   0.25   0.32   0.36   0.39   0.46   3535    1.0\n",
       "y_pred[10]    0.21  8.6e-4   0.05    0.1   0.17   0.21   0.24   0.31   3943    1.0\n",
       "y_pred[11]    0.35  8.3e-4   0.05   0.24   0.31   0.35   0.38   0.45   4000    1.0\n",
       "y_pred[12]    0.17  8.5e-4   0.05   0.07   0.13   0.17   0.21   0.28   4000    1.0\n",
       "y_pred[13]     0.3  8.4e-4   0.05    0.2   0.27    0.3   0.34    0.4   4000    1.0\n",
       "y_pred[14]     0.3  8.6e-4   0.05   0.19   0.27    0.3   0.34   0.41   3869    1.0\n",
       "y_pred[15]    0.14  8.5e-4   0.05   0.03    0.1   0.14   0.18   0.25   4000    1.0\n",
       "y_pred[16]    0.31  8.5e-4   0.05   0.21   0.27   0.31   0.34   0.41   3864    1.0\n",
       "y_pred[17]    0.26  8.7e-4   0.05   0.15   0.22   0.26   0.29   0.36   3802    1.0\n",
       "y_pred[18]    0.42  9.7e-4   0.06   0.31   0.39   0.43   0.46   0.53   3333    1.0\n",
       "y_pred[19]    0.23  8.9e-4   0.05   0.13    0.2   0.23   0.27   0.34   3665    1.0\n",
       "y_pred[20]    0.12  9.3e-4   0.05 9.5e-3   0.08   0.12   0.15   0.23   3510    1.0\n",
       "y_pred[21]    0.15  8.5e-4   0.05   0.05   0.12   0.15   0.19   0.26   4000    1.0\n",
       "y_pred[22]    0.15  8.5e-4   0.05   0.05   0.12   0.15   0.19   0.26   3922    1.0\n",
       "y_pred[23]    0.21  8.6e-4   0.05    0.1   0.18   0.21   0.25   0.32   3834    1.0\n",
       "y_pred[24]    0.17  8.7e-4   0.05   0.07   0.13   0.17    0.2   0.28   3838    1.0\n",
       "y_pred[25]    0.19  8.9e-4   0.05   0.08   0.15   0.19   0.22   0.29   3821    1.0\n",
       "y_pred[26]    0.32  8.4e-4   0.05   0.21   0.28   0.32   0.35   0.42   4000    1.0\n",
       "y_pred[27]    0.32  8.3e-4   0.05   0.21   0.28   0.31   0.35   0.42   4000    1.0\n",
       "y_pred[28]    0.38  9.2e-4   0.05   0.28   0.35   0.39   0.42   0.49   3467    1.0\n",
       "y_pred[29]    0.31  8.4e-4   0.05    0.2   0.27   0.31   0.34   0.41   4000    1.0\n",
       "y_pred[30]    0.25  9.0e-4   0.05   0.14   0.21   0.25   0.28   0.35   3546    1.0\n",
       "y_pred[31]     0.1  8.8e-4   0.06  -0.01   0.06    0.1   0.14   0.21   4000    1.0\n",
       "y_pred[32]     0.2  8.7e-4   0.05    0.1   0.17    0.2   0.24   0.31   3880    1.0\n",
       "y_pred[33]    0.18  8.4e-4   0.05   0.08   0.15   0.18   0.22   0.29   3986    1.0\n",
       "y_pred[34]    0.33  8.7e-4   0.05   0.22   0.29   0.33   0.36   0.43   3677    1.0\n",
       "y_pred[35]    0.34  8.4e-4   0.05   0.23    0.3   0.34   0.37   0.44   3885    1.0\n",
       "y_pred[36]    0.15  8.4e-4   0.05   0.04   0.12   0.15   0.18   0.25   3974    1.0\n",
       "y_pred[37]     0.3  8.4e-4   0.05    0.2   0.27    0.3   0.34   0.41   4000    1.0\n",
       "y_pred[38]    0.27  8.6e-4   0.05   0.16   0.23   0.27    0.3   0.38   4000    1.0\n",
       "y_pred[39]    0.27  9.4e-4   0.05   0.16   0.23   0.27    0.3   0.37   3312    1.0\n",
       "y_pred[40]    0.33  8.3e-4   0.05   0.23    0.3   0.33   0.37   0.43   4000    1.0\n",
       "y_pred[41]    0.34  8.3e-4   0.05   0.23   0.31   0.34   0.38   0.45   4000    1.0\n",
       "y_pred[42]    0.32  8.4e-4   0.05   0.22   0.29   0.32   0.36   0.43   4000    1.0\n",
       "y_pred[43]    0.36  8.5e-4   0.05   0.26   0.33   0.36    0.4   0.47   4000    1.0\n",
       "y_pred[44]    0.42  9.3e-4   0.05   0.31   0.38   0.42   0.46   0.53   3470    1.0\n",
       "y_pred[45]    0.29  8.3e-4   0.05   0.19   0.26   0.29   0.33    0.4   4000    1.0\n",
       "y_pred[46]    0.21  9.8e-4   0.06   0.09   0.17   0.21   0.24   0.32   3407    1.0\n",
       "y_pred[47]    0.37  8.5e-4   0.05   0.26   0.33   0.37    0.4   0.47   3972    1.0\n",
       "y_pred[48]    0.28  8.6e-4   0.05   0.18   0.25   0.29   0.32   0.39   3695    1.0\n",
       "y_pred[49]    0.14  8.7e-4   0.05   0.04    0.1   0.14   0.18   0.25   3734    1.0\n",
       "epsilon[0] -5.2e-4  8.4e-4   0.05  -0.11  -0.04 2.4e-4   0.04    0.1   4000    1.0\n",
       "epsilon[1] -1.3e-3  8.3e-4   0.05   -0.1  -0.04-5.5e-4   0.03    0.1   4000    1.0\n",
       "epsilon[2]  1.9e-4  8.2e-4   0.05   -0.1  -0.04 5.0e-4   0.04    0.1   4000    1.0\n",
       "epsilon[3] -7.6e-4  8.2e-4   0.05   -0.1  -0.04 8.7e-4   0.03    0.1   4000    1.0\n",
       "epsilon[4]  6.3e-4  8.6e-4   0.05   -0.1  -0.03 6.1e-4   0.04    0.1   3747    1.0\n",
       "epsilon[5] -2.4e-4  8.5e-4   0.05   -0.1  -0.03 1.0e-3   0.03    0.1   3660    1.0\n",
       "epsilon[6]  6.1e-4  8.4e-4   0.05   -0.1  -0.03 1.2e-3   0.04    0.1   3832    1.0\n",
       "epsilon[7] -4.6e-4  8.3e-4   0.05   -0.1  -0.03 1.4e-4   0.03    0.1   3868    1.0\n",
       "epsilon[8] -5.1e-4  8.2e-4   0.05   -0.1  -0.04-5.9e-4   0.03    0.1   3881    1.0\n",
       "epsilon[9] -6.6e-4  8.7e-4   0.05   -0.1  -0.03-1.5e-3   0.03    0.1   3581    1.0\n",
       "epsilon[10]-1.8e-3  8.4e-4   0.05   -0.1  -0.04-1.4e-3   0.03    0.1   3942    1.0\n",
       "epsilon[11] 1.1e-4  8.2e-4   0.05   -0.1  -0.03-3.6e-4   0.03    0.1   4000    1.0\n",
       "epsilon[12]-9.9e-4  8.3e-4   0.05   -0.1  -0.04-1.8e-3   0.03    0.1   4000    1.0\n",
       "epsilon[13]-6.7e-4  8.2e-4   0.05  -0.11  -0.03-7.2e-4   0.03    0.1   4000    1.0\n",
       "epsilon[14]-4.3e-4  8.4e-4   0.05  -0.11  -0.03-3.0e-4   0.03    0.1   3953    1.0\n",
       "epsilon[15]-4.2e-4  8.2e-4   0.05  -0.11  -0.03 5.5e-4   0.03    0.1   4000    1.0\n",
       "epsilon[16]-2.6e-4  8.2e-4   0.05   -0.1  -0.04 1.6e-4   0.03    0.1   4000    1.0\n",
       "epsilon[17] 5.1e-4  8.2e-4   0.05   -0.1  -0.03-5.8e-4   0.04    0.1   4000    1.0\n",
       "epsilon[18] 3.1e-4  8.3e-4   0.05   -0.1  -0.03 4.6e-4   0.03   0.11   4000    1.0\n",
       "epsilon[19] 5.6e-4  8.3e-4   0.05   -0.1  -0.03 4.5e-4   0.03   0.11   3853    1.0\n",
       "epsilon[20]-1.2e-4  8.3e-4   0.05  -0.11  -0.04 3.1e-4   0.04    0.1   4000    1.0\n",
       "epsilon[21]-1.5e-3  8.3e-4   0.05  -0.11  -0.04-4.4e-4   0.03    0.1   4000    1.0\n",
       "epsilon[22]-4.5e-4  8.2e-4   0.05   -0.1  -0.04-9.7e-5   0.03    0.1   3993    1.0\n",
       "epsilon[23]-4.7e-4  8.2e-4   0.05   -0.1  -0.04-8.0e-5   0.03    0.1   3975    1.0\n",
       "epsilon[24] 1.7e-4  8.4e-4   0.05   -0.1  -0.03 1.3e-3   0.03   0.11   3888    1.0\n",
       "epsilon[25]-8.6e-4  8.4e-4   0.05  -0.11  -0.03-2.3e-5   0.03    0.1   4000    1.0\n",
       "epsilon[26] 4.1e-4  8.3e-4   0.05   -0.1  -0.03-3.2e-4   0.03   0.11   4000    1.0\n",
       "epsilon[27]-9.2e-4  8.2e-4   0.05   -0.1  -0.03-1.3e-3   0.03    0.1   4000    1.0\n",
       "epsilon[28]-3.3e-4  8.7e-4   0.05   -0.1  -0.04 1.1e-4   0.03    0.1   3613    1.0\n",
       "epsilon[29]-2.2e-4  8.3e-4   0.05   -0.1  -0.03-1.0e-4   0.03    0.1   4000    1.0\n",
       "epsilon[30]-1.7e-3  8.4e-4   0.05   -0.1  -0.04-7.8e-4   0.03    0.1   3761    1.0\n",
       "epsilon[31] 1.5e-3  8.4e-4   0.05   -0.1  -0.03 2.2e-3   0.04    0.1   4000    1.0\n",
       "epsilon[32]-4.6e-4  8.4e-4   0.05  -0.11  -0.04 2.8e-4   0.04    0.1   3940    1.0\n",
       "epsilon[33] 7.6e-4  8.2e-4   0.05   -0.1  -0.03-4.3e-4   0.04    0.1   4000    1.0\n",
       "epsilon[34]-2.4e-4  8.4e-4   0.05   -0.1  -0.03 5.9e-4   0.03    0.1   3765    1.0\n",
       "epsilon[35]-1.0e-3  8.2e-4   0.05   -0.1  -0.03-1.4e-3   0.03    0.1   3922    1.0\n",
       "epsilon[36]-5.8e-4  8.2e-4   0.05   -0.1  -0.03-1.3e-3   0.03    0.1   4000    1.0\n",
       "epsilon[37]-3.5e-5  8.3e-4   0.05   -0.1  -0.03-5.8e-5   0.03   0.11   4000    1.0\n",
       "epsilon[38]-3.6e-5  8.3e-4   0.05  -0.11  -0.03 8.3e-4   0.03    0.1   4000    1.0\n",
       "epsilon[39]-5.5e-4  8.4e-4   0.05  -0.11  -0.03-8.1e-4   0.03    0.1   3908    1.0\n",
       "epsilon[40]-1.2e-3  8.2e-4   0.05   -0.1  -0.04-2.0e-3   0.03    0.1   4000    1.0\n",
       "epsilon[41] 5.7e-4  8.1e-4   0.05   -0.1  -0.03 4.1e-4   0.03    0.1   4000    1.0\n",
       "epsilon[42]-2.0e-4  8.3e-4   0.05   -0.1  -0.04 1.1e-4   0.04    0.1   4000    1.0\n",
       "epsilon[43]-2.5e-4  8.3e-4   0.05   -0.1  -0.04-1.5e-3   0.03    0.1   4000    1.0\n",
       "epsilon[44] 4.5e-4  8.4e-4   0.05   -0.1  -0.03 3.5e-4   0.04    0.1   3813    1.0\n",
       "epsilon[45] 3.4e-4  8.1e-4   0.05   -0.1  -0.03 4.6e-4   0.03    0.1   4000    1.0\n",
       "epsilon[46]-8.4e-4  8.5e-4   0.05  -0.11  -0.04-4.0e-4   0.04    0.1   3843    1.0\n",
       "epsilon[47]-6.3e-4  8.2e-4   0.05   -0.1  -0.04-1.1e-3   0.03    0.1   4000    1.0\n",
       "epsilon[48] 4.8e-4  8.1e-4   0.05   -0.1  -0.03 9.4e-4   0.04    0.1   4000    1.0\n",
       "epsilon[49] 6.5e-4  8.2e-4   0.05   -0.1  -0.03-7.4e-4   0.04    0.1   3945    1.0\n",
       "lp__         120.8    0.04    1.5 116.93 120.05 121.12  121.9 122.68   1416    1.0\n",
       "\n",
       "Samples were drawn using NUTS at Thu May  3 07:03:26 2018.\n",
       "For each parameter, n_eff is a crude measure of effective sample size,\n",
       "and Rhat is the potential scale reduction factor on split chains (at \n",
       "convergence, Rhat=1)."
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_code=\"\"\"\n",
    "data {\n",
    "  int N;\n",
    "  int<lower=0, upper=1> A[N];\n",
    "  real<lower=0, upper=1> Score[N];\n",
    "  real<lower=0, upper=1> Y[N];\n",
    "}\n",
    "parameters {\n",
    "  real b1;\n",
    "  real b2;\n",
    "  real b3;\n",
    "  real<lower=0> sigma;\n",
    "}\n",
    "transformed parameters {\n",
    "  real mu[N];\n",
    "  for (n in 1:N)\n",
    "    mu[n] = b1 + b2*A[n] + b3*Score[n];\n",
    "}\n",
    "model {\n",
    "  for (n in 1:N)\n",
    "    Y[n] ~ normal(mu[n], sigma);\n",
    "}\n",
    "generated quantities {\n",
    "  real y_pred[N];\n",
    "  real epsilon[N];\n",
    "  for (n in 1:N) {\n",
    "    y_pred[n] = normal_rng(mu[n], sigma);\n",
    "    epsilon[n] = y_pred[n] - mu[n];  \n",
    "  }\n",
    "}\n",
    "\"\"\"\n",
    "\n",
    "from pystan import StanModel\n",
    "data = dict(N=df.shape[0], A=df.A, Score=df.Score/200, Y=df.Y)\n",
    "sm = StanModel(model_code=model_code)\n",
    "\n",
    "fit = sm.sampling(data=data, n_jobs=1, seed=123)\n",
    "fit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(4000, 50)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ms = fit.extract()\n",
    "ms['epsilon'].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xd097208>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWkAAAD4CAYAAAAuNhccAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCyGEjf1/CZgLlESeuBsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xe995f98>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 全サンプリングを区別せずにプロット\n",
    "sns.distplot(ms['epsilon'].flatten())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWkAAAD3CAYAAADfYKXJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nbZAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xfadf320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 全サンプリングをIDごとに区別してプロット\n",
    "for i in range(ms['epsilon'].shape[1]):\n",
    "    sns.distplot(ms['epsilon'][:,i], hist=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0xe5f3c88>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD3CAYAAADi8sSvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xfe50240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# サンプリングの平均値のヒストグラム\n",
    "sns.distplot(ms['epsilon'].mean(axis=0))"
   ]
  }
 ],
 "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": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}