Ch08-ex03.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>X</th>\n",
       "      <th>Y</th>\n",
       "      <th>KID</th>\n",
       "      <th>GID</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>7</td>\n",
       "      <td>457</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10</td>\n",
       "      <td>482</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>16</td>\n",
       "      <td>518</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>25</td>\n",
       "      <td>535</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>427</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    X    Y  KID  GID\n",
       "0   7  457    1    1\n",
       "1  10  482    1    1\n",
       "2  16  518    1    1\n",
       "3  25  535    1    1\n",
       "4   5  427    1    1"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('chap08/input/data-salary-3.txt')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## メカニズム１：a, b の平均は業界によって異なるが、a, b のバラツキは業界を問わず一定とする\n",
    "\n",
    "$ a_{avg}[g] $ ~ $ Normal(a_{avg\\_all}, \\sigma_{ag}) $\n",
    "\n",
    "$ b_{avg}[g] $ ~ $ Normal(b_{avg\\_all}, \\sigma_{bg}) $\n",
    "\n",
    "$a[k]$ ~ $ Normal(a_{avg}[GID[g]], \\sigma_a)$\n",
    "\n",
    "$b[k]$ ~ $ Normal(b_{avg}[GID[g]], \\sigma_b)$\n",
    "\n",
    "$Y[n]$ ~ $ Normal( a[KID[n]] + b[KID[n]] X[n], \\sigma_y )$\n",
    "\n",
    "$g = 1, ..., G$\n",
    "\n",
    "$k = 1, ..., K$\n",
    "\n",
    "$n = 1, ..., N$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:pystan:COMPILING THE C++ CODE FOR MODEL anon_model_b8866c67f9f8304959b334edb363f1c3 NOW.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Inference for Stan model: anon_model_b8866c67f9f8304959b334edb363f1c3.\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",
       "aavg_all 311.87   50.09 812.27 -788.2 289.52 384.53 473.78 1164.3    263   1.01\n",
       "bavg_all  20.59    4.37  83.74  -54.2  10.85  18.41  25.38 121.64    368    1.0\n",
       "aavg[0]  360.37     0.8  29.11 303.37  340.7 360.68 379.85 417.68   1334    1.0\n",
       "aavg[1]  298.73    0.48  14.13 270.15 289.67 298.86  307.8 326.66    880    1.0\n",
       "aavg[2]  498.78    1.05  28.75 441.66 479.56 498.98  518.1 554.61    747    1.0\n",
       "bavg[0]   13.09    0.06   2.74   7.68  11.25  13.02  14.93  18.59   2297    1.0\n",
       "bavg[1]    28.6    0.03   1.41  25.72  27.68  28.59   29.5  31.43   1814    1.0\n",
       "bavg[2]   12.53    0.07   2.47   7.67   10.9  12.59  14.14  17.39   1115    1.0\n",
       "a[0]     365.72    0.69   26.7 315.01  347.4 366.42 384.08 418.56   1507    1.0\n",
       "a[1]     357.98    0.76  29.04 298.99 338.62 357.93  377.3 414.93   1466    1.0\n",
       "a[2]     352.84    0.96  36.45 278.25 330.95 354.11 376.28 422.73   1439    1.0\n",
       "a[3]     364.96    0.97  39.59 288.17 339.22 364.17 389.91 445.51   1668    1.0\n",
       "a[4]     293.31    0.58  24.66 241.63 277.84 294.16  308.7 341.25   1826    1.0\n",
       "a[5]     327.02    1.58  33.96 273.07 302.84 322.25 347.14 407.72    464   1.01\n",
       "a[6]     308.48    0.61  20.44  271.5  293.9 307.44 321.72 351.95   1106    1.0\n",
       "a[7]     300.15    0.65  29.28  242.3 282.52 299.16 317.63 361.63   2003    1.0\n",
       "a[8]     287.07    0.67  26.08 230.18 272.05 288.83 304.47 334.24   1533    1.0\n",
       "a[9]     273.23    0.83  21.26 229.61 260.02 274.53 288.09  312.0    655    1.0\n",
       "a[10]    297.79    0.69  28.86 238.75 281.15 297.75 314.39 358.82   1732    1.0\n",
       "a[11]    294.72    0.58  25.51 240.67 279.31 294.45 310.25  346.8   1903    1.0\n",
       "a[12]    320.42    1.04  27.52 274.38 300.45 317.41  337.5 381.84    700    1.0\n",
       "a[13]    289.09    0.49  20.53 246.03 276.17 290.17 302.44 327.34   1791    1.0\n",
       "a[14]    295.28    0.57  25.35 245.71 280.26 295.74 310.91 345.63   1970    1.0\n",
       "a[15]    292.58    0.69  28.74  231.3  275.5 293.77 310.51 347.62   1758    1.0\n",
       "a[16]    295.57    0.62  26.67 238.97 279.42 296.12 312.33 348.13   1843    1.0\n",
       "a[17]    304.15    0.69  28.43 249.37 286.46 303.24 321.19 365.57   1718    1.0\n",
       "a[18]    277.03    1.18  32.91 199.41 259.11 281.65 299.03 330.32    773    1.0\n",
       "a[19]    305.11    0.65  26.01 255.66 288.31 303.26 321.25 360.86   1583    1.0\n",
       "a[20]    290.08    0.58  24.53 238.05 275.25 291.82 306.27 336.04   1773    1.0\n",
       "a[21]    323.18     1.3  27.16 277.17 303.36 320.71 340.25 382.31    434    1.0\n",
       "a[22]    498.83    1.06   35.6 425.67 475.35 498.96 521.29 569.86   1123    1.0\n",
       "a[23]    503.19    0.99   33.0 438.07 480.84 503.39  525.2 569.25   1113    1.0\n",
       "a[24]     505.7    1.01  33.39 442.68 483.04 505.74 527.12 573.56   1083    1.0\n",
       "a[25]     497.1     1.1  35.39 427.24 473.96 497.56 520.92 566.18   1032    1.0\n",
       "a[26]    500.22    1.02  33.04 433.36 478.51 500.71 522.37 563.24   1053    1.0\n",
       "a[27]    497.57    1.18  37.35 421.79 473.26 498.86 522.36 568.31    996    1.0\n",
       "a[28]    494.64     1.1  34.79 425.56 471.49 495.07 518.89 560.48    995    1.0\n",
       "a[29]    495.85    1.23  38.93 417.45 471.63 496.83 521.78 571.14   1007    1.0\n",
       "b[0]       8.87    0.04   1.75   5.51   7.68   8.87  10.02  12.24   1627    1.0\n",
       "b[1]      17.53    0.06   2.32  12.98  15.97  17.49   19.1  21.99   1663    1.0\n",
       "b[2]      10.71    0.05   1.98   6.87   9.41   10.7  11.92  14.84   1690    1.0\n",
       "b[3]      14.34    0.05   2.09  10.27  12.94  14.36  15.77  18.43   1885    1.0\n",
       "b[4]      33.17    0.05   2.09  29.07  31.81  33.12  34.56  37.53   1843    1.0\n",
       "b[5]      36.17    0.11   2.88  29.91  34.47  36.37  38.12  41.35    708    1.0\n",
       "b[6]      30.88    0.03   1.29  28.27  30.04  30.89  31.77  33.28   1771    1.0\n",
       "b[7]      25.22    0.04   1.73  21.76  24.09  25.25  26.34  28.64   1947    1.0\n",
       "b[8]      22.41    0.04   1.45  19.67  21.43  22.39  23.34  25.43   1565    1.0\n",
       "b[9]      29.09    0.06   1.79  25.74  27.86  29.02  30.29  32.79    835    1.0\n",
       "b[10]     32.53    0.04   1.74  29.01  31.39  32.54  33.64  35.96   1921    1.0\n",
       "b[11]     34.19    0.03   1.36  31.42  33.32  34.17  35.04  36.91   2317    1.0\n",
       "b[12]     28.81    0.07   2.27  24.19  27.35  28.89  30.33  33.14   1196    1.0\n",
       "b[13]      24.0    0.04    2.0  20.11  22.65  23.98  25.33  27.88   2301    1.0\n",
       "b[14]     33.77    0.06   2.79   28.5  31.89  33.68  35.52  39.43   2126    1.0\n",
       "b[15]     19.47    0.03   1.38  16.81  18.55  19.48  20.36  22.25   1962    1.0\n",
       "b[16]     28.21    0.03   1.37  25.56  27.27  28.21  29.14  30.91   2459    1.0\n",
       "b[17]      28.9    0.05   2.33   24.1  27.43  28.91  30.43  33.31   2031    1.0\n",
       "b[18]     20.82    0.07   2.35  16.67  19.25  20.63  22.25  25.93   1119    1.0\n",
       "b[19]      34.0    0.03   1.54   31.0  33.01  34.03  35.01  37.01   1935    1.0\n",
       "b[20]     26.59    0.05   2.45  21.74  24.98   26.5  28.21  31.46   2180    1.0\n",
       "b[21]      28.2    0.05   1.34  25.38  27.33  28.28  29.12  30.65    693    1.0\n",
       "b[22]      13.1    0.05   1.94   9.28  11.79  13.08  14.43  16.94   1731    1.0\n",
       "b[23]     13.87    0.06   2.08   9.77  12.51  13.88  15.29  17.95   1344    1.0\n",
       "b[24]     13.41    0.07   2.38   8.69  11.83  13.42  14.97  18.08   1308    1.0\n",
       "b[25]     11.92    0.09   3.29   5.54   9.77  11.89  14.14  18.34   1201    1.0\n",
       "b[26]     12.69    0.08   3.16   6.44  10.59  12.63  14.81  18.92   1483    1.0\n",
       "b[27]     11.72    0.08   2.79   6.41   9.86  11.65   13.6  17.19   1259    1.0\n",
       "b[28]     11.49    0.11   4.06   3.71   8.65   11.5   14.3  19.39   1462    1.0\n",
       "b[29]     11.46    0.06    2.1   7.13  10.07  11.44  12.86  15.56   1115    1.0\n",
       "s_ag     654.71  123.54 2254.4  62.87  128.7 222.64 482.16 3716.6    333   1.01\n",
       "s_bg      69.56   18.97 540.27   5.74   11.4  18.92  37.88 352.29    811   1.01\n",
       "s_a        25.8     0.9   12.3   5.27  16.74  24.91  33.63   51.8    185   1.01\n",
       "s_b        4.75    0.02   0.81   3.38   4.19   4.66   5.21   6.59   2587    1.0\n",
       "s_y       65.27    0.05   2.91  59.97  63.26  65.12  67.26  71.15   4000    1.0\n",
       "lp__      -1582    1.24  15.49  -1606  -1592  -1584  -1574  -1543    156   1.01\n",
       "\n",
       "Samples were drawn using NUTS at Fri May  4 14:58:34 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": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_code=\"\"\"\n",
    "data {\n",
    "  int G;\n",
    "  int K;\n",
    "  int N;\n",
    "  real<lower=0> X[N];\n",
    "  real<lower=0> Y[N];\n",
    "  int<lower=1,upper=G> K2G[K];\n",
    "  int<lower=1,upper=K> KID[N];\n",
    "}\n",
    "parameters {\n",
    "  real aavg_all;\n",
    "  real bavg_all;\n",
    "  real aavg[G];\n",
    "  real bavg[G];\n",
    "  real a[K];\n",
    "  real b[K];\n",
    "  real<lower=0> s_ag;\n",
    "  real<lower=0> s_bg;\n",
    "  real<lower=0> s_a;\n",
    "  real<lower=0> s_b;\n",
    "  real<lower=0> s_y;\n",
    "}\n",
    "model {\n",
    "  for (g in 1:G) {\n",
    "    aavg[g] ~ normal(aavg_all, s_ag);\n",
    "    bavg[g] ~ normal(bavg_all, s_bg);\n",
    "  }\n",
    "  for (k in 1:K) {\n",
    "    a[k] ~ normal(aavg[K2G[k]], s_a);\n",
    "    b[k] ~ normal(bavg[K2G[k]], s_b);  \n",
    "  }\n",
    "\n",
    "  for (n in 1:N)\n",
    "    Y[n] ~ normal(a[KID[n]]+b[KID[n]]*X[n], s_y);\n",
    "}\n",
    "\"\"\"\n",
    "\n",
    "from pystan import StanModel\n",
    "\n",
    "K2G = df[~df[['KID', 'GID']].duplicated()].GID\n",
    "data = dict(N=df.shape[0], G=df.GID.nunique(), K=df.KID.nunique(), X=df.X, K2G=K2G, KID=df.KID, Y=df.Y)\n",
    "sm = StanModel(model_code=model_code)\n",
    "\n",
    "fit = sm.sampling(data=data, n_jobs=1, seed=1234)\n",
    "fit"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 業界平均 $a_{avg}[g]$ と $b_{avg}[g]$ の事後分布を作図する"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\eX\\Anaconda3\\lib\\site-packages\\seaborn\\categorical.py:588: FutureWarning: remove_na is deprecated and is a private function. Do not use.\n",
      "  kde_data = remove_na(group_data)\n",
      "C:\\Users\\eX\\Anaconda3\\lib\\site-packages\\seaborn\\categorical.py:816: FutureWarning: remove_na is deprecated and is a private function. Do not use.\n",
      "  violin_data = remove_na(group_data)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEFCAYAAAABjYvXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngD8Bk4mFotzkO7gAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xf76cac8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ms = fit.extract()\n",
    "\n",
    "fig = plt.figure()\n",
    "ax=plt.subplot(121)\n",
    "plt.xlabel('GID')\n",
    "plt.ylabel('a_avg')\n",
    "sns.violinplot(data=ms['aavg'], color=sns.palettes.color_palette()[0])\n",
    "\n",
    "\n",
    "ax=plt.subplot(122)\n",
    "sns.violinplot(data=ms['bavg'], color=sns.palettes.color_palette()[0])\n",
    "plt.xlabel('GID')\n",
    "plt.ylabel('b_avg')\n",
    "\n",
    "plt.show()"
   ]
  }
 ],
 "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
}