{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Missing values and resampling\n",
    "\n",
    "Allen Downey\n",
    "\n",
    "[MIT License](https://en.wikipedia.org/wiki/MIT_License)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "import statsmodels.formula.api as smf"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Missing value imputation\n",
    "\n",
    "This notebook demonstrates a version of [\"hot-deck\" missing value imputation](https://en.wikipedia.org/wiki/Imputation_(statistics)).\n",
    "\n",
    "First I'll load a dataset from the General Social Survey (GSS)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the data file\n",
    "\n",
    "import os\n",
    "\n",
    "datafile = 'gss_eda.hdf5'\n",
    "if not os.path.exists(datafile):\n",
    "    !wget https://github.com/AllenDowney/PoliticalAlignmentCaseStudy/raw/master/gss_eda.hdf5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64814, 165)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss = pd.read_hdf(datafile, 'gss')\n",
    "gss.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It includes 64814 respondents and 165 variables for each respondent."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Suppose we want to run a model that compares income for men and women, controlling for age and education level.\n",
    "\n",
    "Here's what it looks like if we don't fill missing data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>OLS Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>         <td>realinc</td>     <th>  R-squared:         </th>  <td>   0.183</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                   <td>OLS</td>       <th>  Adj. R-squared:    </th>  <td>   0.183</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>             <td>Least Squares</td>  <th>  F-statistic:       </th>  <td>   2605.</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>             <td>Fri, 17 Apr 2020</td> <th>  Prob (F-statistic):</th>   <td>  0.00</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                 <td>12:12:35</td>     <th>  Log-Likelihood:    </th> <td>-6.7458e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>No. Observations:</th>      <td> 58110</td>      <th>  AIC:               </th>  <td>1.349e+06</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Residuals:</th>          <td> 58104</td>      <th>  BIC:               </th>  <td>1.349e+06</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Model:</th>              <td>     5</td>      <th>                     </th>      <td> </td>     \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>      <td>nonrobust</td>    <th>                     </th>      <td> </td>     \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "       <td></td>          <th>coef</th>     <th>std err</th>      <th>t</th>      <th>P>|t|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>   <td>-2.452e+04</td> <td> 1384.380</td> <td>  -17.713</td> <td> 0.000</td> <td>-2.72e+04</td> <td>-2.18e+04</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(sex)[T.2]</th> <td>-4612.6446</td> <td>  222.704</td> <td>  -20.712</td> <td> 0.000</td> <td>-5049.145</td> <td>-4176.144</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>educ</th>        <td> -294.8175</td> <td>  168.502</td> <td>   -1.750</td> <td> 0.080</td> <td> -625.082</td> <td>   35.447</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>educ2</th>       <td>  143.3503</td> <td>    6.622</td> <td>   21.648</td> <td> 0.000</td> <td>  130.371</td> <td>  156.329</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>age</th>         <td> 1698.3989</td> <td>   36.532</td> <td>   46.491</td> <td> 0.000</td> <td> 1626.797</td> <td> 1770.001</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>age2</th>        <td>  -16.9586</td> <td>    0.365</td> <td>  -46.436</td> <td> 0.000</td> <td>  -17.674</td> <td>  -16.243</td>\n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "  <th>Omnibus:</th>       <td>20760.569</td> <th>  Durbin-Watson:     </th> <td>   1.627</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Prob(Omnibus):</th>  <td> 0.000</td>   <th>  Jarque-Bera (JB):  </th> <td>74398.723</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Skew:</th>           <td> 1.807</td>   <th>  Prob(JB):          </th> <td>    0.00</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Kurtosis:</th>       <td> 7.203</td>   <th>  Cond. No.          </th> <td>3.68e+04</td> \n",
       "</tr>\n",
       "</table><br/><br/>Warnings:<br/>[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.<br/>[2] The condition number is large, 3.68e+04. This might indicate that there are<br/>strong multicollinearity or other numerical problems."
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                            OLS Regression Results                            \n",
       "==============================================================================\n",
       "Dep. Variable:                realinc   R-squared:                       0.183\n",
       "Model:                            OLS   Adj. R-squared:                  0.183\n",
       "Method:                 Least Squares   F-statistic:                     2605.\n",
       "Date:                Fri, 17 Apr 2020   Prob (F-statistic):               0.00\n",
       "Time:                        12:12:35   Log-Likelihood:            -6.7458e+05\n",
       "No. Observations:               58110   AIC:                         1.349e+06\n",
       "Df Residuals:                   58104   BIC:                         1.349e+06\n",
       "Df Model:                           5                                         \n",
       "Covariance Type:            nonrobust                                         \n",
       "===============================================================================\n",
       "                  coef    std err          t      P>|t|      [0.025      0.975]\n",
       "-------------------------------------------------------------------------------\n",
       "Intercept   -2.452e+04   1384.380    -17.713      0.000   -2.72e+04   -2.18e+04\n",
       "C(sex)[T.2] -4612.6446    222.704    -20.712      0.000   -5049.145   -4176.144\n",
       "educ         -294.8175    168.502     -1.750      0.080    -625.082      35.447\n",
       "educ2         143.3503      6.622     21.648      0.000     130.371     156.329\n",
       "age          1698.3989     36.532     46.491      0.000    1626.797    1770.001\n",
       "age2          -16.9586      0.365    -46.436      0.000     -17.674     -16.243\n",
       "==============================================================================\n",
       "Omnibus:                    20760.569   Durbin-Watson:                   1.627\n",
       "Prob(Omnibus):                  0.000   Jarque-Bera (JB):            74398.723\n",
       "Skew:                           1.807   Prob(JB):                         0.00\n",
       "Kurtosis:                       7.203   Cond. No.                     3.68e+04\n",
       "==============================================================================\n",
       "\n",
       "Warnings:\n",
       "[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.\n",
       "[2] The condition number is large, 3.68e+04. This might indicate that there are\n",
       "strong multicollinearity or other numerical problems.\n",
       "\"\"\""
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['age2'] = gss['age']**2\n",
    "gss['educ2'] = gss['educ']**2\n",
    "\n",
    "model = smf.ols('realinc ~ educ + educ2 + age + age2 + C(sex)', data=gss)\n",
    "results = model.fit()\n",
    "results.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that the number of observations is 58110.  The model has quietly dropped respondents who are missing any of the independents variables or the dependent variable.\n",
    "\n",
    "We can plot the results like this."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_results(results):\n",
    "    df = pd.DataFrame()\n",
    "    df['age'] = np.linspace(18, 89)\n",
    "    df['educ'] = 16\n",
    "\n",
    "    df['age2'] = df['age']**2\n",
    "    df['educ2'] = df['educ']**2\n",
    "\n",
    "    df['sex'] = 1\n",
    "    pred1 = results.predict(df)\n",
    "    pred1.index = df['age']\n",
    "\n",
    "    df['sex'] = 2\n",
    "    pred2 = results.predict(df)\n",
    "    pred2.index = df['age']\n",
    "    \n",
    "    pred1.plot(label='Male', alpha=0.6)\n",
    "    pred2.plot(label='Female', alpha=0.6)\n",
    "\n",
    "    plt.xlabel('Age')\n",
    "    plt.ylabel('Real income (1986 $)')\n",
    "    plt.title('Real income versus age, grouped by sex')\n",
    "    plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "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": [
    "plot_results(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see where those missing values are."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "228"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['age'].isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are 228 respondents with unknown age."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "177"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['educ'].isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are 177 respondents with unknown education level."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['sex'].isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are no respondents with unknown sex."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6521"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['realinc'].isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And there are 6521 respondents with unknown income.  So let's start filling things in."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following function takes a Pandas Series and modifies it in place, replacing any NaN values with randomly-chosen valid values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fill_missing_values(series):\n",
    "    \"\"\"Fills missing values of the given Series.\n",
    "\n",
    "    series: Pandas Series\n",
    "    \"\"\"\n",
    "    # Boolean Series: true where values are missing\n",
    "    null = series.isnull()\n",
    "\n",
    "    # Series of valid values\n",
    "    valid = series.dropna()\n",
    "    \n",
    "    # the fill values are a sample of the valid values \n",
    "    fill = valid.sample(null.sum(), replace=True)\n",
    "    \n",
    "    # we need the index of the fill values to line up\n",
    "    # with the index of the DataFrame\n",
    "    fill.index = series.index[null]\n",
    "\n",
    "    # replace NaNs with the fill values\n",
    "    series.fillna(fill, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, I will fill missing values for `age` and `educ`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "varnames = ['age', 'educ']\n",
    "for varname in varnames:\n",
    "    fill_missing_values(gss[varname])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And we can confirm that there are no NaNs left."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['age'].isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['educ'].isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I will not bother to fill `realinc` because it is the dependent variable.\n",
    "\n",
    "Filling in the independent variables make it possible to include respondents with partial information, so it helps by adding information to the model.\n",
    "\n",
    "Filling in the dependent variable would do nothing but add randomness to the results.  If the missing values introduce sampling bias (for example, if people with high income are less likely to respond), filling missing values would not help with that problem.\n",
    "\n",
    "Now I'll run the model again with the filled data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>OLS Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>         <td>realinc</td>     <th>  R-squared:         </th>  <td>   0.183</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                   <td>OLS</td>       <th>  Adj. R-squared:    </th>  <td>   0.183</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>             <td>Least Squares</td>  <th>  F-statistic:       </th>  <td>   2605.</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>             <td>Fri, 17 Apr 2020</td> <th>  Prob (F-statistic):</th>   <td>  0.00</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                 <td>12:49:36</td>     <th>  Log-Likelihood:    </th> <td>-6.7674e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>No. Observations:</th>      <td> 58293</td>      <th>  AIC:               </th>  <td>1.353e+06</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Residuals:</th>          <td> 58287</td>      <th>  BIC:               </th>  <td>1.354e+06</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Model:</th>              <td>     5</td>      <th>                     </th>      <td> </td>     \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>      <td>nonrobust</td>    <th>                     </th>      <td> </td>     \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "       <td></td>          <th>coef</th>     <th>std err</th>      <th>t</th>      <th>P>|t|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>   <td>-2.451e+04</td> <td> 1381.328</td> <td>  -17.743</td> <td> 0.000</td> <td>-2.72e+04</td> <td>-2.18e+04</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(sex)[T.2]</th> <td>-4636.0289</td> <td>  222.482</td> <td>  -20.838</td> <td> 0.000</td> <td>-5072.095</td> <td>-4199.962</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>educ</th>        <td> -280.3947</td> <td>  168.059</td> <td>   -1.668</td> <td> 0.095</td> <td> -609.791</td> <td>   49.002</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>educ2</th>       <td>  142.6127</td> <td>    6.605</td> <td>   21.590</td> <td> 0.000</td> <td>  129.666</td> <td>  155.559</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>age</th>         <td> 1695.1496</td> <td>   36.486</td> <td>   46.461</td> <td> 0.000</td> <td> 1623.637</td> <td> 1766.662</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>age2</th>        <td>  -16.9267</td> <td>    0.365</td> <td>  -46.415</td> <td> 0.000</td> <td>  -17.641</td> <td>  -16.212</td>\n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "  <th>Omnibus:</th>       <td>20842.655</td> <th>  Durbin-Watson:     </th> <td>   1.628</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Prob(Omnibus):</th>  <td> 0.000</td>   <th>  Jarque-Bera (JB):  </th> <td>74760.189</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Skew:</th>           <td> 1.808</td>   <th>  Prob(JB):          </th> <td>    0.00</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Kurtosis:</th>       <td> 7.207</td>   <th>  Cond. No.          </th> <td>3.68e+04</td> \n",
       "</tr>\n",
       "</table><br/><br/>Warnings:<br/>[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.<br/>[2] The condition number is large, 3.68e+04. This might indicate that there are<br/>strong multicollinearity or other numerical problems."
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                            OLS Regression Results                            \n",
       "==============================================================================\n",
       "Dep. Variable:                realinc   R-squared:                       0.183\n",
       "Model:                            OLS   Adj. R-squared:                  0.183\n",
       "Method:                 Least Squares   F-statistic:                     2605.\n",
       "Date:                Fri, 17 Apr 2020   Prob (F-statistic):               0.00\n",
       "Time:                        12:49:36   Log-Likelihood:            -6.7674e+05\n",
       "No. Observations:               58293   AIC:                         1.353e+06\n",
       "Df Residuals:                   58287   BIC:                         1.354e+06\n",
       "Df Model:                           5                                         \n",
       "Covariance Type:            nonrobust                                         \n",
       "===============================================================================\n",
       "                  coef    std err          t      P>|t|      [0.025      0.975]\n",
       "-------------------------------------------------------------------------------\n",
       "Intercept   -2.451e+04   1381.328    -17.743      0.000   -2.72e+04   -2.18e+04\n",
       "C(sex)[T.2] -4636.0289    222.482    -20.838      0.000   -5072.095   -4199.962\n",
       "educ         -280.3947    168.059     -1.668      0.095    -609.791      49.002\n",
       "educ2         142.6127      6.605     21.590      0.000     129.666     155.559\n",
       "age          1695.1496     36.486     46.461      0.000    1623.637    1766.662\n",
       "age2          -16.9267      0.365    -46.415      0.000     -17.641     -16.212\n",
       "==============================================================================\n",
       "Omnibus:                    20842.655   Durbin-Watson:                   1.628\n",
       "Prob(Omnibus):                  0.000   Jarque-Bera (JB):            74760.189\n",
       "Skew:                           1.808   Prob(JB):                         0.00\n",
       "Kurtosis:                       7.207   Cond. No.                     3.68e+04\n",
       "==============================================================================\n",
       "\n",
       "Warnings:\n",
       "[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.\n",
       "[2] The condition number is large, 3.68e+04. This might indicate that there are\n",
       "strong multicollinearity or other numerical problems.\n",
       "\"\"\""
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gss['age2'] = gss['age']**2\n",
    "gss['educ2'] = gss['educ']**2\n",
    "\n",
    "model2 = smf.ols('realinc ~ educ + educ2 + age + age2 + C(sex)', data=gss)\n",
    "results2 = model2.fit()\n",
    "results2.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And plot the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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": [
    "plot_results(results)\n",
    "plot_results(results2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is no visible difference between the results with and without filling, which suggests that for this dataset missing data is not a substantial source of uncertainty.\n",
    "\n",
    "If it were, I would consider other ways of imputing missing values, including [regression](https://en.wikipedia.org/wiki/Imputation_(statistics)#Regression)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Resampling\n",
    "\n",
    "Next we'll use resampling to quantify variability in the model due to random sampling.  The following function takes a DataFrame and resamples the rows."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def resample_rows_weighted(df, column):\n",
    "    \"\"\"Resamples a DataFrame using probabilities proportional to given column.\n",
    "\n",
    "    df: DataFrame\n",
    "    column: string column name to use as weights\n",
    "\n",
    "    returns: DataFrame\n",
    "    \"\"\"\n",
    "    weights = df[column]\n",
    "    sample = df.sample(n=len(df), replace=True, weights=weights)\n",
    "    return sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The GSS uses stratified sampling, which means that some groups are deliberately oversampled.  The variable `wtssall` contains sampling weights, which we use during resampling to undersample the groups that were oversampled, and vice versa.\n",
    "\n",
    "As a result, in the resampled dataset, every row has the same sampling weight; that is, the resampled dataset is representative of the population.\n",
    "\n",
    "In the GSS dataset, the sampling weights are computed for each cycle of data collection, so I'll group the dataset by year, resample within each year, and then put the results back into one DataFrame."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "def resample_by_year(df, column):\n",
    "    \"\"\"Resample rows within each year.\n",
    "\n",
    "    df: DataFrame\n",
    "    column: string name of weight variable\n",
    "\n",
    "    returns DataFrame\n",
    "    \"\"\"\n",
    "    grouped = df.groupby('year')\n",
    "    samples = [resample_rows_weighted(group, column)\n",
    "               for _, group in grouped]\n",
    "    sample = pd.concat(samples, ignore_index=True)\n",
    "    return sample"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here's how we use these functions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "gss = pd.read_hdf(datafile, 'gss')\n",
    "sample = resample_by_year(gss, 'wtssall')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following function takes a resampled dataset, runs the model, and returns the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_model(df):\n",
    "    varnames = ['age', 'educ']\n",
    "    for varname in varnames:\n",
    "        fill_missing_values(df[varname])\n",
    "    \n",
    "    df['age2'] = df['age']**2\n",
    "    df['educ2'] = df['educ']**2\n",
    "\n",
    "    model = smf.ols('realinc ~ educ + educ2 + age + age2 + C(sex)', data=df)\n",
    "    results = model.fit()\n",
    "    return results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here's how we can use it with one resampled dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>OLS Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>         <td>realinc</td>     <th>  R-squared:         </th>  <td>   0.172</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                   <td>OLS</td>       <th>  Adj. R-squared:    </th>  <td>   0.172</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>             <td>Least Squares</td>  <th>  F-statistic:       </th>  <td>   2396.</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>             <td>Fri, 17 Apr 2020</td> <th>  Prob (F-statistic):</th>   <td>  0.00</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                 <td>12:55:14</td>     <th>  Log-Likelihood:    </th> <td>-6.7331e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>No. Observations:</th>      <td> 57712</td>      <th>  AIC:               </th>  <td>1.347e+06</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Residuals:</th>          <td> 57706</td>      <th>  BIC:               </th>  <td>1.347e+06</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Model:</th>              <td>     5</td>      <th>                     </th>      <td> </td>     \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>      <td>nonrobust</td>    <th>                     </th>      <td> </td>     \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "       <td></td>          <th>coef</th>     <th>std err</th>      <th>t</th>      <th>P>|t|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>   <td>-2.335e+04</td> <td> 1453.046</td> <td>  -16.070</td> <td> 0.000</td> <td>-2.62e+04</td> <td>-2.05e+04</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(sex)[T.2]</th> <td>-4010.7110</td> <td>  235.822</td> <td>  -17.007</td> <td> 0.000</td> <td>-4472.924</td> <td>-3548.498</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>educ</th>        <td>  -72.5272</td> <td>  178.890</td> <td>   -0.405</td> <td> 0.685</td> <td> -423.152</td> <td>  278.097</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>educ2</th>       <td>  143.2837</td> <td>    7.059</td> <td>   20.299</td> <td> 0.000</td> <td>  129.448</td> <td>  157.119</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>age</th>         <td> 1652.1489</td> <td>   38.586</td> <td>   42.817</td> <td> 0.000</td> <td> 1576.520</td> <td> 1727.778</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>age2</th>        <td>  -16.5914</td> <td>    0.396</td> <td>  -41.845</td> <td> 0.000</td> <td>  -17.369</td> <td>  -15.814</td>\n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "  <th>Omnibus:</th>       <td>17926.666</td> <th>  Durbin-Watson:     </th> <td>   1.990</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Prob(Omnibus):</th>  <td> 0.000</td>   <th>  Jarque-Bera (JB):  </th> <td>52014.279</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Skew:</th>           <td> 1.645</td>   <th>  Prob(JB):          </th> <td>    0.00</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Kurtosis:</th>       <td> 6.286</td>   <th>  Cond. No.          </th> <td>3.43e+04</td> \n",
       "</tr>\n",
       "</table><br/><br/>Warnings:<br/>[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.<br/>[2] The condition number is large, 3.43e+04. This might indicate that there are<br/>strong multicollinearity or other numerical problems."
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                            OLS Regression Results                            \n",
       "==============================================================================\n",
       "Dep. Variable:                realinc   R-squared:                       0.172\n",
       "Model:                            OLS   Adj. R-squared:                  0.172\n",
       "Method:                 Least Squares   F-statistic:                     2396.\n",
       "Date:                Fri, 17 Apr 2020   Prob (F-statistic):               0.00\n",
       "Time:                        12:55:14   Log-Likelihood:            -6.7331e+05\n",
       "No. Observations:               57712   AIC:                         1.347e+06\n",
       "Df Residuals:                   57706   BIC:                         1.347e+06\n",
       "Df Model:                           5                                         \n",
       "Covariance Type:            nonrobust                                         \n",
       "===============================================================================\n",
       "                  coef    std err          t      P>|t|      [0.025      0.975]\n",
       "-------------------------------------------------------------------------------\n",
       "Intercept   -2.335e+04   1453.046    -16.070      0.000   -2.62e+04   -2.05e+04\n",
       "C(sex)[T.2] -4010.7110    235.822    -17.007      0.000   -4472.924   -3548.498\n",
       "educ          -72.5272    178.890     -0.405      0.685    -423.152     278.097\n",
       "educ2         143.2837      7.059     20.299      0.000     129.448     157.119\n",
       "age          1652.1489     38.586     42.817      0.000    1576.520    1727.778\n",
       "age2          -16.5914      0.396    -41.845      0.000     -17.369     -15.814\n",
       "==============================================================================\n",
       "Omnibus:                    17926.666   Durbin-Watson:                   1.990\n",
       "Prob(Omnibus):                  0.000   Jarque-Bera (JB):            52014.279\n",
       "Skew:                           1.645   Prob(JB):                         0.00\n",
       "Kurtosis:                       6.286   Cond. No.                     3.43e+04\n",
       "==============================================================================\n",
       "\n",
       "Warnings:\n",
       "[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.\n",
       "[2] The condition number is large, 3.43e+04. This might indicate that there are\n",
       "strong multicollinearity or other numerical problems.\n",
       "\"\"\""
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results3 = run_model(sample)\n",
    "results3.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we can compare the results with the original and resampled datasets."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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": [
    "plot_results(results)\n",
    "plot_results(results3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this example, there is a substantial difference, which suggests that the groups that were oversampled tended to have lower incomes.  When we use resampling to correct for stratified sampling, predicted incomes increase."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Confidence intervals\n",
    "\n",
    "We can use repeated resampling to generate confidence intervals.  The following look resamples the dataset 11 times, runs the model, and collects the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_seq = []\n",
    "gss = pd.read_hdf(datafile, 'gss')\n",
    "\n",
    "for i in range(11):\n",
    "    sample = resample_by_year(gss, 'wtssall')\n",
    "    results = run_model(sample)\n",
    "    result_seq.append(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I'll use the following function to generate predictions for each resampled dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "def make_predict(results, sex):\n",
    "    df = pd.DataFrame()\n",
    "    df['age'] = np.linspace(18, 89)\n",
    "    df['educ'] = 16\n",
    "\n",
    "    df['age2'] = df['age']**2\n",
    "    df['educ2'] = df['educ']**2\n",
    "\n",
    "    df['sex'] = sex\n",
    "    pred = results.predict(df)\n",
    "    pred.index = df['age']\n",
    "    return pred"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A quick way to visualize variability due to sampling is to plot predictions for each of the resampled dataset using a low value of `alpha`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "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": [
    "for results in result_seq:\n",
    "    pred = make_predict(results, 1)\n",
    "    pred.plot(color='C0', alpha=0.1)\n",
    "\n",
    "for results in result_seq:\n",
    "    pred = make_predict(results, 2)\n",
    "    pred.plot(color='C1', alpha=0.1)\n",
    "    \n",
    "plt.xlabel('Age')\n",
    "plt.ylabel('Real income (1986 $)')\n",
    "plt.title('Real income versus age, grouped by sex');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And alternative is to collect the predictions in a list:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_seq = []\n",
    "gss = pd.read_hdf(datafile, 'gss')\n",
    "\n",
    "for i in range(51):\n",
    "    sample = resample_by_year(gss, 'wtssall')\n",
    "    results = run_model(sample)\n",
    "    result_seq.append(results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The result is a list of Series, but we can treat it as an array with one row for each resampled dataset and one column for each of the ages in the range.\n",
    "\n",
    "Then we can use `np.percentiles` to compute the 5th, 50th, and 95th percentile in each columns (`axis=0` means we perform the computation along the first axis, which is down the columns).\n",
    "\n",
    "The result is an array with 3 rows, which we can assign to variables.  Then we use `fill_between` to plot the area between the 5th and 95th percentiles, and `plot` to draw a line through the median values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_percentiles(result_seq, sex, **options):\n",
    "    pred_list = []\n",
    "\n",
    "    for results in result_seq:\n",
    "        pred = make_predict(results, sex)\n",
    "        pred_list.append(pred)\n",
    "\n",
    "    low, med, high = np.percentile(pred_list, [5, 50, 95], axis=0)\n",
    "    plt.fill_between(pred.index, low, high, alpha=0.3)\n",
    "    plt.plot(pred.index, med, **options)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here's what it looks like."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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": [
    "plot_percentiles(result_seq, 1, label='Male')\n",
    "plot_percentiles(result_seq, 2, label='Female')\n",
    "\n",
    "plt.xlabel('Age')\n",
    "plt.ylabel('Real income (1986 $)')\n",
    "plt.title('Real income versus age, grouped by sex')\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The 90% confidence intervals are barely wider than the plotted lines, which suggests that random sampling is not a big source of uncertainty.\n",
    "\n",
    "In summary, for this example:\n",
    "\n",
    "1. Missing values are not a substantial source of uncertainty, so we could use a simple form of imputation, or (even simpler) drop rows with missing values.\n",
    "\n",
    "2. Stratified sampling has a big effect on the results, so it is important to correct for it.\n",
    "\n",
    "3. The confidence intervals show that random sampling contributes only small variations in the model's predictions."
   ]
  }
 ],
 "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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}