{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Why Causal Analysis is Needed\n",
    "In this example we will perform a quick causal analysis to estimate the _causal effect_ of smoking cessation on weight gain over a period of a decade.   \n",
    "We will compare these results to the _observed effect_, arising from simple descriptive statistics.  \n",
    "We'll then dig deeper into the data and see why such discrepancies arise."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The Data\n",
    "We will use data from an NHEFS observational study. \n",
    "The NHANS (National Health and Nutrition Examionation Survey) Epidemiologic Followup Study (NHEFS) is a national longitudinal study that was jointly initiated by the American National Center for Health Statistics and the American National Institute on Aging. \n",
    "The study was designed to follow up on smokers, some of whom quit smoking, and record their weight (among other measurements) for a period of 11 years: from 1971 to 1982.  \n",
    "More information about the data can be found in the [NHEFS Notebook](/notebooks/nhefs.ipynb)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "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>age</th>\n",
       "      <th>race</th>\n",
       "      <th>sex</th>\n",
       "      <th>smokeintensity</th>\n",
       "      <th>smokeyrs</th>\n",
       "      <th>wt71</th>\n",
       "      <th>active_1</th>\n",
       "      <th>active_2</th>\n",
       "      <th>education_2</th>\n",
       "      <th>education_3</th>\n",
       "      <th>education_4</th>\n",
       "      <th>education_5</th>\n",
       "      <th>exercise_1</th>\n",
       "      <th>exercise_2</th>\n",
       "      <th>age^2</th>\n",
       "      <th>wt71^2</th>\n",
       "      <th>smokeintensity^2</th>\n",
       "      <th>smokeyrs^2</th>\n",
       "      <th>qsmk</th>\n",
       "      <th>wt82_71</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>42</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>30</td>\n",
       "      <td>29</td>\n",
       "      <td>79.04</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1764</td>\n",
       "      <td>6247.3216</td>\n",
       "      <td>900</td>\n",
       "      <td>841</td>\n",
       "      <td>0</td>\n",
       "      <td>-10.093960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>36</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>24</td>\n",
       "      <td>58.63</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1296</td>\n",
       "      <td>3437.4769</td>\n",
       "      <td>400</td>\n",
       "      <td>576</td>\n",
       "      <td>0</td>\n",
       "      <td>2.604970</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>56</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>20</td>\n",
       "      <td>26</td>\n",
       "      <td>56.81</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3136</td>\n",
       "      <td>3227.3761</td>\n",
       "      <td>400</td>\n",
       "      <td>676</td>\n",
       "      <td>0</td>\n",
       "      <td>9.414486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>68</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>53</td>\n",
       "      <td>59.42</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>4624</td>\n",
       "      <td>3530.7364</td>\n",
       "      <td>9</td>\n",
       "      <td>2809</td>\n",
       "      <td>0</td>\n",
       "      <td>4.990117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>40</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>19</td>\n",
       "      <td>87.09</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1600</td>\n",
       "      <td>7584.6681</td>\n",
       "      <td>400</td>\n",
       "      <td>361</td>\n",
       "      <td>0</td>\n",
       "      <td>4.989251</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   age  race  sex  ...  smokeyrs^2  qsmk    wt82_71\n",
       "0   42     1    0  ...         841     0 -10.093960\n",
       "1   36     0    0  ...         576     0   2.604970\n",
       "2   56     1    1  ...         676     0   9.414486\n",
       "3   68     1    0  ...        2809     0   4.990117\n",
       "4   40     0    0  ...         361     0   4.989251\n",
       "\n",
       "[5 rows x 20 columns]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from causallib.datasets import load_nhefs\n",
    "\n",
    "data = load_nhefs()\n",
    "data.X.join(data.a).join(data.y).head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The data is already divided into three components: \n",
    " * `X`: the confounders needed to be controlled for:\n",
    "   * Demographics: such as `age`, `race`, `sex` and `education` level (how many years of schools).\n",
    "   * Smoking habits: how many years an individual has been smoking (`smokeyrs`) and how many cigarettes per day (`smokeintensity`).\n",
    "   * Well-being habits: how much active a participant is daily (`active`) and how much exercise they do in recreation (`exercise`).\n",
    "   * Weight at the start of the study (`wt71`).\n",
    " * `a`: the treatment assignment - whether an individual had quit smoking or not.\n",
    " * `y`: the outcome - how much weight the individual gained during the follow up period of a decade."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Estimating Causal Effect\n",
    "### The Causal Model\n",
    "For this demosntation, we will follow the most simple and well known causal model: _Inverse Probability of Treatment Weighting_ (also known as IPTW or IPW).  \n",
    "This model requires estimating the propensity of each participant to stop smoking based on their individual features.   \n",
    "Modeling this $\\Pr[A=1|X]$ requires a machine learning model.  \n",
    "`causallib` allows users to plug in any arbitrary ML model they want. \n",
    "For the sake of simplicity, we will use a logistic regression model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from causallib.estimation import IPW\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "learner = LogisticRegression(penalty='none',  # No regularization, new in scikit-learn 0.21.*\n",
    "                             solver='lbfgs',\n",
    "                             max_iter=500)    # Increased to achieve convergence with 'lbfgs' solver\n",
    "ipw = IPW(learner)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Applying Causal Analysis\n",
    "Once we defined the causal model, we can move on to calculate the potential outcomes and the effect"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.5175400213088306"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ipw.fit(data.X, data.a)\n",
    "potential_outcomes = ipw.estimate_population_outcome(data.X, data.a, data.y)\n",
    "causal_effect = potential_outcomes[1] - potential_outcomes[0]\n",
    "causal_effect"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our causal analysis concluded that, on average, smoking cessation contributed 3.5 kg (7.7 lbs) to the weight gain of people over a period of a decade.  \n",
    "Put differently, had **all** the participants stoped smoking, they would have weighed 3.5 kg more than if **all** participants had continued to smoke."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Estimating Observed Effect\n",
    "Say you are tasked with estimating how much smoking cessation contributes to weight gain but you are not familiar with causal analysis.  \n",
    "On the face of it, it might sound simple. You simply measure the weight-gain in the people who continued smoking, you measure the weight gain in people who quit, and compare the two quantities to see how much the people who quit gained more (or less) than the people who persisted.  \n",
    "Basically you measure $\\mathbb{E}[Y|A=1] - \\mathbb{E}[Y|A=0]$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.540581454955888"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "observed_outcomes = data.y.groupby(data.a).mean()\n",
    "observed_effect = observed_outcomes[1] - observed_outcomes[0]\n",
    "observed_effect"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So if you were to avoid causal analysis, you would estimate that smoking cessation contributes only 2.5 kg (5.5 lbs) to the weight gain, as opposed to the 3.5 kg we calculated above. This is an under-estimation of close to 30 (!) percent.  \n",
    "What can cause such a discrepancy between the descriptive-statistics calculation and the causal-based estimation?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Analysis of the difference between causal and observed effect\n",
    "We saw that the observed difference due to smoking cessation is 2.5 kg, while the causal inference analysis suggested it is closer to 3.5.  \n",
    "Causal analysis, in its essence, tries to balance between the treatment and control groups before calculating the difference.  \n",
    "Weight gain can be caused by many factors. \n",
    "For example, males tend to be larger than females on average, and they gain more weight on average. What if most of the people who quit smoking were female? That would bring down the observed effect.\n",
    "Our observed analysis didn't account for the different distributions of such features between the treatment groups, and therefore the observed effect is _confounded_ and most likely \"contaminated\" with _spuriuous correlations_. \n",
    "Only when controlling for such correlations and eliminating their effect, we can claim that the differences in weight gain are truly only due to the status of smoking.\n",
    "\n",
    "What the observed difference indicates is only that people who **chose** to quit smoking were observed to gain 2.5 Kg more than people who **chose** not to quit smoking. \n",
    "\n",
    "To get a clue of how the distribution differ between the treated and the controls, we can calculate the difference in means between groups for each variables. \n",
    "We visualize these differences in a plot called Love plot (named after Dr. Thomas Love)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from causallib.evaluation import evaluate\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "evaluations = evaluate(ipw, data.X, data.a, data.y, metrics_to_evaluate={})\n",
    "fig, ax = plt.subplots(figsize=(6, 6))\n",
    "evaluations.plot_covariate_balance(kind=\"love\", ax=ax);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The orange triangles show the differences in averages between for each covariates _before_ applying balancing.  \n",
    "We see that `age` differs the most between the treatment groups, and we suspect it might be causing the discrepancy in results between the descriptive statistics and the causal analysis.\n",
    "\n",
    "To further investigate this, let's look at the weight-gain as a function of the age:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/micha/causal/lib/python3.8/site-packages/seaborn-0.11.0-py3.8.egg/seaborn/_decorators.py:36: FutureWarning: Pass the following variables as keyword args: x, y. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 648x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "\n",
    "ax = sns.scatterplot(data.X[\"age\"], data.y, hue=data.a, alpha=0.7)\n",
    "ax.get_figure().set_size_inches(9, 5);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We observe that for older age people the weight-gain decreases to the point that it is more frequently negative than positive at very high age.\n",
    "\n",
    "Let's break down the observed difference by age groups of 5 years each:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0    (40.0, 45.0]\n",
       "1    (35.0, 40.0]\n",
       "2    (55.0, 60.0]\n",
       "3    (65.0, 70.0]\n",
       "4    (35.0, 40.0]\n",
       "Name: age, dtype: category\n",
       "Categories (10, interval[float64, right]): [(24.999, 30.0] < (30.0, 35.0] < (35.0, 40.0] < (40.0, 45.0] ... (55.0, 60.0] < (60.0, 65.0] < (65.0, 70.0] < (70.0, 75.0]]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "step_size = 5\n",
    "age_groups = range(data.X[\"age\"].min(), \n",
    "                   data.X[\"age\"].max() + step_size, \n",
    "                   step_size)\n",
    "print(list(age_groups))\n",
    "age_grouped = pd.cut(data.X[\"age\"], bins=age_groups, include_lowest=True)\n",
    "age_grouped.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now we examine how the weight gain differ in each age-group:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "observed_diff_by_age = data.y.groupby([data.a, age_grouped]).mean()\n",
    "observed_diff_by_age = observed_diff_by_age.xs(1) - observed_diff_by_age.xs(0)\n",
    "observed_diff_by_age = observed_diff_by_age.rename(\"observed_effect\")\n",
    "ax = observed_diff_by_age.plot(kind=\"barh\")\n",
    "ax.set_xlabel(\"observed_effect\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see here two interseting trends:\n",
    " * The oldest age group has an observed effect much lower than the others.\n",
    " * The people between ages 40-55 have larger observed difference than the average 2.5 kg.\n",
    "It means when we do a simple average across all the treated and controls, we are mixing people with different observed effects:\n",
    "$$\n",
    "\\mathbb{E}\\left[Y|A=1,\\text{Age}\\gt70\\right] - \\mathbb{E}\\left[Y|A=0,\\text{Age}\\gt70\\right] \\ll \n",
    "\\mathbb{E}\\left[Y|A=1,\\text{Age}\\in[40,55]\\right] - \\mathbb{E}\\left[Y|A=0,\\text{Age}\\in[40,55]\\right]\n",
    "$$ \n",
    "\n",
    "This certainly has to be adjusted for. But how much? the overall effect will depend in the number of people in each age group:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAasAAAEGCAYAAADbk7pdAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAAmZUlEQVR4nO3df5RdZX3v8feHQRLQ8jP+gIAmCF4Ff0Q8UmyxIr0VBBdQixqbC6GoXCu0t2FJG1ZvrQt1FaretNEWF0gEXLSBSxFTCb8N1HVbhAkGkqBICqlkoFIJCdJgLPFz/9jPyGY4JzOTnD3nZPJ5rXXW7P3s59eeE+bLfvaz9yPbRERE9LNdet2BiIiI0SRYRURE30uwioiIvpdgFRERfS/BKiIi+t6uve7AZDRt2jTPmDGj192IiNihLF++/Ce2X97uWIJVA2bMmMHg4GCvuxERsUOR9G+djmUYMCIi+l6CVURE9L0Eq4iI6HsJVhER0fcywaIBK4c2MmP+DeMut/bCExvoTUTEji9XVhER0fcSrCIiou81Eqwk7S7pTkkDkt4taUXt8zNJp5R8MyV9V9IaSVdL2q1DfeeXPA9KOm4M7X+n1t5jkq4v6cdI2lg79qkO5dv2S9I8ST+S9OVt/d1ERMT4NXVldSZwne0ttpfZnmV7FnAssAm4peS7CFhg+xDgKeAjIyuSdBgwGzgcOB74W0kDW2vc9jtrbf4LcF3t8HeGj9m+oEMVbftlewHQNsBFRERzmgpWc4Bvtkk/FbjR9iZJogpe15ZjVwCntClzMrDY9mbbjwBrgCPH0glJe5Y2rh9rx8fRr5HlzpI0KGlwy6aNY20uIiLGoOvBqgyZHWx7bZvDs4G/L9v7ARtsP1f21wHT25SZDjxa2++Ur51TgNttP11Le4ek+yTdKOnwNmXG2q8XsH2J7Zbt1sAee42xexERMRZNTF2fBmwYmShpf+BNwM0NtNnJh4Gv1vbvBV5j+xlJJ1BdcR06gf2JiIht0MQw4LPA1DbpHwS+Yfu/yv6TwN6ShgPmgcBQm3JDwEG1/U75XkDSNKrhwl8+8GT7advPlO2lwEtKvrqx9isiIiZI14OV7aeAAUkjA9aHeX4IENsGllHdxwKYS/v7XEuA2ZKmSJpJdSV0N4Ck2yV1GqI7FfiW7Z8NJ0h6VbknhaQjqc7/yRH9H2u/IiJigjQ1weIW4OjhHUkzqK6O7hyR70+AcyWtobpXdFnJf5KkCwBsrwauAR4AbgLOtr1F0i7AIcD6Dn2o3x8bdiqwStJ9wEJgdglOSFoq6YCt9SsiInpD5W91dyuVjgDm2T6t65U/38YbgTNtn9tUGx3aPQNo2T6nU55Wq+WsZxURMT6SlttutTvWyJWV7XuBZaM9D7WdbazqQaCaB5wPPD1a3oiI6J7GXmRre1FTdfdKeSh4Qa/7ERGxs8m7ASMiou8lWEVERN9LsIqIiL6XYBUREX0vwSoiIvpeglVERPS9BKuIiOh7CVYREdH3GnsoeGe2cmgjM+bfMHrGSWTthSf2ugsRMYnlyioiIvpeglVERPS9RoKVpN0l3Tn8IltJr5Z0i6TvS3qgLBmCpMslPSJpRfnM6lDfXEkPlc/cMbR/da3OtZJW1I6dL2mNpAclHdeh/ExJ3y35rpa0W0mfJ+lHkr487l9KRERss6buWZ0JXGd7S9m/Evic7VslvQz4RS3vebav7VSRpH2BPwdagIHlkpaURR7bsv2hWvkvAhvL9mFU61wdDhwA3CbpdbV+DrsIWGB7saSvAB8BLra9QNJTpS8RETFBmhoGnENZXbcEiF1t3wpg+xnbm8ZR13HArbbXlwB1K3D8WAqWVYE/yPOLMJ4MLLa92fYjwBrgyDZljgWGA+gVwCljaOssSYOSBrds2jiW7kVExBh1PViVIbODba8tSa8DNki6TtL3JH1+xDpXn5N0v6QFkqa0qXI68Ghtf11JG4t3Aj+2/dA46toP2GD7ufG0Z/sS2y3brYE99hpj9yIiYiyauLKaBmyo7e9KFTQ+CbwdOBg4oxw7H3h9Sd+Xajn5bvowL17aPiIidjBNBKtngam1/XXACtsPl6uV64EjAGw/7spm4GuMGJIrhoCDavsHlrStkrQr8H7g6nHW9SSwdyk/5vYiIqI5XQ9W5b7SgKThgHUP1R//l5f9Y4EHACTtX36K6r7QqjZV3gy8R9I+kvYB3lPSkHSlpHYBDuC/Az+wva6WtgSYLWmKpJnAocDdI/pvYBlwakmaS7n/FhERvdHUbMBbgKOB22xvkfRJ4PYSlJYDl5Z8V5UgJmAF8HEASS3g47Y/anu9pM9QBT2AC2yvL9tvBh7r0IfZjBgCtL1a0jVUwfI54OzhmYCSlgIftf0Y1XDkYkmfBb4HXDaek3/T9L0YzBsdIiK6RtWFRJcrlY4A5tk+reuVP9/GnsBltj/QVBsd2j0DaNk+p1OeVqvlwcHBietURMQkIGm57baPBjUydd32vcCyEbP+ut3G0z0IVPOoJoU8PZHtRkTs7Bp7ka3tRU3V3Su2FwALet2PiIidTd4NGBERfS/BKiIi+l6CVURE9L0Eq4iI6HsJVhER0fcSrCIiou8lWEVERN9r7DmrndnKoY3MmH9Dr7ux01ubV15FTBq5soqIiL6XYBUREX2vkWAlaXdJdw6/G1DSqyXdIun7kh6QNKOkz5T0XUlrJF1dVhluV9/5Jc+Dko4bQ/uS9DlJPyxt/mFJP0bSRkkryudTHcq37ZekeZJ+JOnL2/iriYiIbdDUldWZwHXDy28AVwKft/0GqgUWnyjpFwELbB8CPAV8ZGRFkg6jWu7jcOB44G/H8ILcM6gWWXx9aXNx7dh3bM8qnws6lG/br/JuwLYBLiIimtNUsJpDWbCwBJtdbd8KYPsZ25vK2lbHAteWMldQLcA40snAYtubbT8CrKH9isJ1v0+17tUvSptPjJL/l8bRr4iImCBdD1ZlyOxg22tL0uuADZKuk/Q9SZ8vV0b7ARvKUvcA64DpbaqcDjxa2++Ur+61wIckDUq6UdKhtWPvkHRfST+8Tdmx9usFJJ1V2hvcsmnjaNkjImIcmriymgZsqO3vCrwT+CTwduBgqmG6Jk0BflYW8boUGF6u5F7gNbbfAnwJuL5bDdq+xHbLdmtgj726VW1ERNBMsHoWmFrbXwessP1wuVq5HjgCeBLYW9Lws14HAkNt6huiuv/EKPnq1gHXle1vAG+GXy7Y+EzZXgq8RNK0EWXH2q+IiJggXQ9Wtp8CBiQNB6x7qP74v7zsHws8YNvAMuDUkj6Xcp9rhCXAbElTJM0EDgXuBpB0u6R2Q3TXA+8u2+8Cfljyv6rck0LSkVTn/+SI/o+1XxERMUGammBxC3A0QJkR+EngdkkrAVENzQH8CXCupDVU94ouA5B0kqQLSvnVwDXAA8BNwNm2t0jaBTgEWN+m/QuB3ynt/QXw0ZJ+KrBK0n3AQmB2CU5IWirpgK31KyIiekPlb3V3K5WOAObZPq3rlT/fxhuBM22f21QbHdo9A2jZPqdTnlar5cHBwYnrVETEJCBpeZlr8CKNXFnZvhdYNobnobanjVU9CFTzgPOBpyey3YiInV1jL7K1vWj0XDuW8lDwgl73IyJiZ5N3A0ZERN9LsIqIiL6XYBUREX0vwSoiIvpeglVERPS9BKuIiOh7CVYREdH3EqwiIqLvNfZQ8M5s5dBGZsy/odfdiEls7YUn9roLERMqV1YREdH3EqwiIqLvNRKsJO0u6c7hF9lK2iJpRfksqeW7XNIjtWOzOtQ3V9JD5TN3jH34A0k/kLRa0l/W0s+XtEbSg5KO61B2pqTvlnxXS9qtpM+T9CNJXx7HryMiIrZTU/eszgSuK2tZATxre1aHvOfZvrZTRZL2Bf4caAEGlktaUhZ57FTm3cDJwFtsb5b0ipJ+GDAbOBw4ALhN0utq/Rx2EbDA9mJJXwE+Alxse4Gkp0pfIiJigjQ1DDiH7q2uexxwq+31JUDdChw/SpnfBy60vRnA9hMl/WRgse3Nth8B1gBH1guWlYSPBYYD6BXAKaN1UtJZkgYlDW7ZtHFsZxYREWPS9WBVhswOtr22ljy1/CG/S9IpI4p8TtL9khZImtKmyunAo7X9dSVta14HvLMM5d0p6e3jqGs/YIPt58bRHrYvsd2y3RrYY6/RskdExDg0cWU1DdgwIu01ZfXH3wX+StJrS/r5wOuBtwP7Ui0n3w27lvqOAs4DrilXTBERsQNqIlg9C0ytJ9geKj8fBu4A3lr2H3dlM/A1RgzJFUPAQbX9A0va1qyjumdm23cDv6AKomOp60lgb0m7biVPRERMoK4Hq3JfaUDSVABJ+wwP70maBvw68EDZ37/8FNV9oVVtqrwZeE+pZx/gPSUNSVdKahfgrgfeXfK8DtgN+AmwBJgtaYqkmcChwN0j+m9gGXBqSZpL9+6/RUTENmhqgsUtwNFl+w3AoKT7qILAhbYfKMeukrQSWEl15fNZAEktSV8FsL0e+AxwT/lcUNIA3gw81qb9RcDBklYBi4G55SprNXANVbC8CTh7eCagpKWSDijl/wQ4V9IaqntYl233byQiIraZqguJLlcqHQHMs31a1yt/vo09gctsf6CpNjq0ewbQsn1OpzytVsuDg4MT16mIiElA0vIyv+FFGrmysn0vsGz4oeCG2ni6B4FqHtWkkKcnst2IiJ1dYy+ytb2oqbp7xfYCYEGv+xERsbPJuwEjIqLvJVhFRETfS7CKiIi+l2AVERF9L8EqIiL6XoJVRET0vQSriIjoewlWERHR9xp7KHhntnJoIzPm39DrbkRMOmsvPLHXXYgeyZVVRET0vQSriIjoe40EK0m7l+XkB8r+FkkrymdJLd/MsvT8GklXS9qtQ33nlzwPSjpuDO1fLumRWpuzSvoxkjbW0j/VoXzbfkmaJ+lHkr68Db+WiIjYRk1dWZ1JtVLvlrL/rO1Z5XNSLd9FwALbhwBPAR8ZWZGkw4DZwOHA8cDfjvFt7ufV2lxRS/9OLf2CDmXb9qu8yLZtgIuIiOY0FazmMMrqumV14GOBa0vSFVSrBY90MrDY9mbbjwBrgHarA3fFOPo1stxZkgYlDW7ZtLGp7kVE7JS6HqzKkNnBttfWkqeWP+R3STqlpO0HbLD9XNlfB0xvU+V04NHafqd8I31O0v2SFkiaUkt/h6T7JN0o6fA25cbarxewfYntlu3WwB57jaF7ERExVk1cWU0DNoxIe01Z/fF3gb+S9NoG2q07H3g98HZgX6pl6gHuLX15C/Al4PqG+xEREV3QRLB6FphaT7A9VH4+DNwBvBV4Ethb0vCzXgcCQ23qGwIOqu13yldv73FXNgNfowwbltWFnynbS4GXSJo2ovhY+xUREROk68HK9lPAgKSpAJL2GR6GK4Hh14EHbBtYBpxais6l/X2uJcBsSVMkzQQOBe4u9d0u6UVDdJL2Lz9Fdb9pVdl/VUlD0pFU5//kiP6PtV8RETFBmppgcQtwdNl+AzAo6T6qIHCh7QfKsT8BzpW0hupe0WUAkk6SdAGA7dXANcADwE3A2ba3SNoFOARY36b9qyStBFZSDUt+tqSfCqwqfVkIzC7BCUlLJR2wtX5FRERvqPyt7m6l0hHAPNundb3y59t4I3Cm7XObaqNDu2cALdvndMrTarU8ODg4cZ2KiJgEJC0v8xtepJErK9v3AsvG+DzUtraxqgeBah7V5I2nJ7LdiIidXWMvsrW9qKm6e6U8FLyg1/2IiNjZ5N2AERHR9xKsIiKi7yVYRURE30uwioiIvpdgFRERfS/BKiIi+l6CVURE9L0Eq4iI6HuNPRS8M1s5tJEZ82/odTciYgez9sITe92FvpUrq4iI6HsJVhER0fcaCVaSdpd0Z/1FtpL2lLRO0pdraXdIelDSivJ5RYf6zpe0puQ9bgztXy7pkVq9s0q6JC0sdd1f3g7frvzbJK0s+RbW1sD6vKR/l/TJcf5KIiJiOzR1z+pM4DrbW2ppnwH+qU3eObY7rqch6TBgNnA4cABwm6TXjai7nfNsXzsi7b1UizceCvwqcHH5OdLFwMeA7wJLgeOBG22fJ+k/R2k3IiK6rKlhwDnUVteV9DbglVSLMo7XycBi25ttPwKsoSxTv411XVmWvL+Lavn6/esZyv6etu8qCzNeSbXa8FZJOkvSoKTBLZs2bmP3IiKina4HK0m7AQfbXlv2dwG+CHQaOvtaGar7s+HhthGmA4/W9teVtNF8rgz1LZA0ZRx1TS/p42rP9iW2W7ZbA3vsNYbuRUTEWDVxZTUN2FDb/wSw1Pa6Nnnn2H4T8M7y6dbKwucDrwfeDuxLtUx9RETsoJoIVs8CU2v77wDOkbQW+AJwuqQLAWwPlZ8/Bf6O9sN7Q8BBtf0DS1pHth8vQ32bga/V6h1LXUMlfcztRUREs7oerGw/BQxImlr259h+te0ZVEOBV9qeL2lXSdMAJL0EeB+wqk2VS4DZkqZImkk1OeLuUu52SS8aohu+D1WGFU+p1buEKlhK0lHARtuPj+j/48DTko4q5U+ndv8tIiImXlOzAW8BjgZu20qeKcDNJVANlLyXAkg6CWjZ/pTt1ZKuAR4AngPOtr2l3As7BFjfpu6rJL0cELAC+HhJXwqcQDVJYxPwe8MFJK2wPavsfgK4HNgduLF8xuxN0/diME+iR0R0jaoJb12utHp+aZ7tbt2DatfGG4EzbZ/bVBsd2v008IztL3TK02q1PDjYcTZ+RES0IWm57Va7Y41MXbd9L7Cs/lBwA22s6kGg+jzwP4A8axURMYEae5Gt7UVN1d0rts8Dzut1PyIidjZ5N2BERPS9BKuIiOh7CVYREdH3EqwiIqLvJVhFRETfS7CKiIi+l2AVERF9r7HnrHZmK4c2MmP+Db3uRkQAa/Pqs0khV1YREdH3EqwiIqLvNRKsJO0u6c76uwEl7SlpnaQv19LeJmmlpDWSFrZbKbgs57Gw5Lm/vCR3rP1YImlVbf/TkobKysQrJJ3Qodzxkh4sbc6vpV8lab2kU8fah4iI2H5NXVmdCVxne0st7TPAP43IdzHwMao1qg4Fjm9T13trx88qZUYl6f3AM20OLbA9q3yWtik3APxNafcw4MOSDoNqbS6qNbEiImICNRWs5lBbsFDS24BXUq1zNZy2P7Cn7btcrVNyJdVCiSOdTLVgo23fBew9vLhiJ5JeBpwLfHYb+n4ksMb2w7Z/DiwufYiIiB7perCStBtwsO21ZX8X4ItUqwTXTQfW1fbXlbSRpgOPjiFf3WdKm5vaHDunDCcukrRPl9pD0lmSBiUNbtm0cbTsERExDk1cWU0DNtT2PwEstb2uffbukjQLeK3tb7Q5fDHwWmAW8DhVQOsK25fYbtluDeyxV7eqjYgImnnO6llgam3/HcA7JX0CeBmwm6RngL8GDqzlOxAYalPfEHDQGPLV22tJWkt1fq+QdIftY2z/eDiTpEuBb3WhvYiIaFjXr6xsPwUMSJpa9ufYfrXtGVRDgVfanm/7ceBpSUeVWYCnU7vPVbMEOL3MCjwK2FjKIukHbdq/2PYBpb2jgR/aPqbkr9/r+m1g1cjywD3AoZJmliHN2WRSRURETzU1weIWqkAxmk8AXwXWAP8K3Agg6eOSPl7yLAUeLnkuLWWQNA140VT3UfxlmSp/P/BuYF6p6wBJSwFsPwecA9wMfB+4xvbqcbYTERFdpGoiXpcrrZ6Fmmf7tK5X/nwb76OayLGwqTY6tHs58C3b13bK02q1PDg4OHGdioiYBCQtt91qd6yRdwPavlfSMkkDI5616mYb7e43NUrSVcCvAR0DVUREdF9jL7K1vaipunulPBQcERETLO8GjIiIvpdgFRERfS/BKiIi+l6CVURE9L0Eq4iI6HsJVhER0fcSrCIiou8lWEVERN9r7KHgndnKoY3MmH9Dr7sRscNYe+GJve5C9LlcWUVERN9LsIqIiL7XSLCStLukOyUNSHqNpHslrZC0urb0B5LukPRgObZC0is61He+pDUl73Hj6MfCstDj8P4USVeXur4raUaHcseXttZIml9Lv0rSekmnjrUPERGx/Zq6Z3UmcJ3tLZIeB95he7OklwGrJC2x/VjJO8d2x/U0JB1GtQDi4cABwG2SXjfa29wltYB9RiR/BHjK9iGSZgMXAR8aUW4A+Bvgt4B1wD2lvw/YnlOWCImIiAnU1DDgHMqqv7Z/bntzSZ+yDW2eDCy2vdn2I1SLMB65tQIl4Hwe+OM2dV1Rtq8FfrOsUlx3JLDG9sO2fw4sLuW2StJZkgYlDW7ZtHHUk4qIiLHrerAqS8EfbHttLe2gsjrvo8BFtasqgK+VIcA/axM4AKaXcsPWlbStOQdYYvvxTnWVFYE3Avt1oT1sX2K7Zbs1sMdeo2WPiIhxaOLKahqwoZ5g+1HbbwYOAeZKemU5NMf2m4B3ls92ryws6QDgA8CXtreuiIjoD00Eq2eBqe0OlCuqVVSBCdtD5edPgb+j/fDeEHBQbf/AktbJW6mC4hpJa4E9JK0ZWZekXYG9gCe3s72IiGhY14OV7aeAAUlTASQdKGn3sr0PcDTwoKRdJU0r6S8B3kcVyEZaAswuM/lmAocCd5dyt0t6wRCd7Rtsv8r2DNszgE22D6nVNbdsnwp827ZHtHcPcKikmWVIc3YpFxERPdLUbMBbqILSbcAbgC9KMiDgC7ZXSnopcHMJVAMl76UAkk4CWrY/ZXu1pGuAB4DngLPLLMNdqK6g1o+jX5cBXy9XWuupAtHw0OFXbZ9g+zlJ5wA3l34tsr16+34dERGxPfTiC4suVCodAcyzvd33oLbSxhuBM22f21QbHdq9HPiW7Ws75Wm1Wh4c7DgbPyIi2pC03Har3bFGpq7bvhdYVqaQN8L2qh4EqquAdwE/m8h2IyJ2do29yNb2oqbq7hXbc3rdh4iInVHeDRgREX0vwSoiIvpeglVERPS9BKuIiOh7CVYREdH3EqwiIqLvJVhFRETfS7CKiIi+19hDwTuzlUMbmTH/hl53IyK6bO2FJ/a6CzutXFlFRETfS7CKiIi+10iwkrS7pDslDUh6jaR7y9L1qyV9vJbvbZJWSlojaWG7Ze1VWVjy3F/e6D5a+zdJuq+095XhF+pK+rSkodKXFZJO6FD+eEkPljbn19KvkrRe0qnb9puJiIht0dSV1ZnAdba3AI8D77A9C/hVYH5ZPwrgYuBjVAsqHgoc36au99aOn1XKjOaDtt8CvBF4OdUy98MW2J5VPktHFiyB7W9Ku4cBH5Z0GPzyRbZZiDEiYoI1FazmAN8EsP1z25tL+pThNiXtD+xp+66yWu+VwClt6joZuNKVu4C9S9mObD9dNncFdgPGs2jXkcAa2w/b/jmwuPRhqySdJWlQ0uCWTRvH0VxERIym68GqLAV/sO21tbSDJN0PPApcZPsxYDqwrlZ0XUkbaXopN1q+kf24GXgC+ClQXyjxnDKcuEjSPt1qz/Yltlu2WwN77DVa9oiIGIcmrqymARvqCbYftf1mqmXo50p6ZQPtvoDt44D9qa7mji3JFwOvBWZRDU9+sel+RETE9msiWD0LTG13oFxRrQLeCQwBB9YOH1jSRhoCDhpDvnbt/YxqOPLksv9j21ts/wK4lGrIr2vtRUREM7oerGw/BQxImgog6UBJu5ftfYCjgQdtPw48LemoMgvwdMp9rhGWAKeXWYFHARtLWST9YGRmSS8bvqclaVfgROAHZb9+r+u3qQLnSPcAh0qaWYY0Z5NJFRERPdXUGyxuoQpKtwFvAL4oyYCAL9heWfJ9Argc2B24sXwYnt5u+yvAUuAEYA2wCfi9kmdaqW+klwJLJA1P5lgGfKUc+0tJs6gmXKwF/mep6wDgq7ZPsP2cpHOAm4EBYJHt1dv9G4mIiG2maiJelyutnoWaZ/u0rlf+fBvvo5rIsbCpNjq0eznwLdvXdsrTarU8ODg4cZ2KiJgEJC233Wp3rJErK9v3SlomaaA8a9VEG99qot6tkXQV8Gu8cHZhREQ0rLEX2dpe1FTdvVIeCo6IiAmWdwNGRETfS7CKiIi+l2AVERF9L8EqIiL6XoJVRET0vQSriIjoewlWERHR9xp7zmpntnJoIzPm39DrbkRETKi1F57YWN25soqIiL6XYBUREX0vwSoiIvreqMFK0u6S7pQ0IGmWpH+RtLosDf+hNvkXSnqmQ127SfqapJWS7pN0TO3Yh0qdqyVdVEt/jaTby7E7JB3Yru4R7dxU6l8t6SuSBkr6vpJulfRQ+dluWXskzS15HpI0t5a+TNIzktq+FTgiIpoxliurM4HrytvTNwGn2z4cOB74K0l7D2csf8TbBoDiYwC23wT8FtU6V7tI2g/4PPCbpe5XSfrNUuYLwJW23wxcAPzFGPr8QdtvAd4IvBz4QEmfD9xu+1Dg9rL/ApL2Bf4c+FWqlYT/fDio2X43kLU/IiIm2FiC1RzKCr62f2j7obL9GPAEVTCgXL18HvjjrdR1GPDtUv4JYAPQAg4GHrL9HyXfbcDvjCxDtZDiyaN12PbTZXNXYDeqxRYpZa8o21cAp7Qpfhxwq+31ZdXjW6kC81ZJOkvSoKTBLZs2jpY9IiLGYavBqizrfrDttW2OHUkVCP61JJ0DLBlecr6D+4CTJO0qaSbwNuAgqlWA/5ukGWUp+lNK+nCZ95ft3wZ+pVyJbZWkm6mC6U95fv2pV9b69+/AK9sUnQ48WttfV9K2yvYltlu2WwN77DVa9oiIGIfRrqymUV39vICk/YGvA79n+xdlWfgPAF8apb5FVH/8B4G/Av4Z2FKuYH4fuBr4DtWS88OLNn4SeJek7wHvAoZqxzqyfRywPzAFOLbNcfP8FVdERPSx0R4KfhaYWk+QtCdwA/Cntu8qyW8FDgHWSALYQ9Ia24fUy9p+DphXq+ufgR+WY/8I/GNJP4sSkMpw4/tL+suA37G9YSwnZ/tnkr5JNfx3K/BjSfvbfrwE3CfaFBsCjqntHwjcMZb2IiKiGVu9sipXPAOSpsIvhwW/QTXh4dpavhtsv8r2DNszgE0jA1Upv4ekl5bt3wKes/1A2X9F+bkP8Angq2V/mqThfp5PdXU2XN8P2rTxshKIKEOKJwLD+ZYAw7P75lLuxY1wM/AeSfuUvrynpEVERI+M5XVLtwBHU016+CDwG8B+ks4ox8+wvaJTYUknAS3bnwJeAdws6RdUVzCn1bL+taS3lO0LbP+wbB8D/IUkA/8EnF3qnQaoTZMvBZZImkIVjJcBXynHLgSukfQR4N/K+QzPYvy47Y/aXi/pM8A9tb6s38rv50XeNH0vBht87UhExM5G1a2brWSQjgDm2T5tqxknmKT3UU3+WDjB7d4BfNJ2xynsrVbLg4OZ4R4RMR6Slttu+xzrqFdWtu8tD8MOlGet+oLtb010m5KWUU2z/6+JbjsiYmc2preu2140eq7JrzwUHBEREyzvBoyIiL6XYBUREX1v1AkWMX6Sfgo82Ot+TIBpwE963YkJsDOc585wjpDz7Hevsf3ydgeyUnAzHuw0o2UykTSY85wcdoZzhJznjizDgBER0fcSrCIiou8lWDXjkl53YILkPCePneEcIee5w8oEi4iI6Hu5soqIiL6XYBUREX0vwarLJB0v6UFJayTN73V/ukXSWkkrJa2QNFjS9pV0q6SHys99et3P8ZK0SNITklbV0tqelyoLy3d7f3nJ8w6hw3l+WtJQ+U5XSDqhduz8cp4PSjquN70eP0kHlXeZPiBptaT/VdInzXe6lXOcdN/nC9jOp0sfYAD4V6qX3e4G3Acc1ut+denc1gLTRqT9JTC/bM8HLup1P7fhvH4DOAJYNdp5AScAN1ItTXMU8N1e9387z/PTVCsIjMx7WPm3OwWYWf5ND/T6HMZ4nvsDR5TtX6Fa3PWwyfSdbuUcJ933Wf/kyqq7jgTW2H7Y9s+BxVSrFE9WJwNXlO0rgFN615VtY/ufgJHrlXU6r5OpFh61q1Wy9x5e6LPfdTjPTk4GFtvebPsRYA3Vv+2+Z/tx2/eW7Z8C3wemM4m+062cYyc77PdZl2DVXdOBR2v769j6P6IdiYFbJC2XdFZJe6Xtx8v2vwOv7E3Xuq7TeU3G7/ecMvy1qDaMOynOU9IM4K3Ad5mk3+mIc4RJ/H0mWMVYHW37COC9wNmSfqN+0NV4w6R7DmKynldxMfBaYBbwOPDFnvamiyS9DPgH4I9sP10/Nlm+0zbnOGm/T0iw6rYh4KDa/oElbYdne6j8fAL4BtUwwo+Hh0zKzyd618Ou6nRek+r7tf1j21ts/wK4lOeHhnbo85T0Eqo/4lfZvq4kT6rvtN05Ttbvc1iCVXfdAxwqaaak3YDZwJIe92m7SXqppF8Z3gbeA6yiOre5Jdtc4Ju96WHXdTqvJcDpZQbZUcDG2tDSDmfEvZnfpvpOoTrP2ZKmSJoJHArcPdH92xaSBFwGfN/2/6kdmjTfaadznIzf5wv0eobHZPtQzS76IdWMmz/tdX+6dE4HU80mug9YPXxewH7A7cBDwG3Avr3u6zac299TDZn8F9VY/kc6nRfVjLG/Kd/tSqDV6/5v53l+vZzH/VR/0Pav5f/Tcp4PAu/tdf/HcZ5HUw3x3Q+sKJ8TJtN3upVznHTfZ/2T1y1FRETfyzBgRET0vQSriIjoewlWERHR9xKsIiKi7yVYRURE30uwiohfkvRHkvbodT8iRsrU9Yj4JUlrqZ41+kmv+xJRlyuriB2MpNPLy0rvk/R1STMkfbuk3S7p1SXf5ZJOrZV7pvw8RtIdkq6V9ANJV5U3OPwhcACwrKyXNFDqWKVqLbN5vTnjCNi11x2IiLGTdDjwv4Ffs/0TSftSLXlxhe0rJJ0JLGT05VreChwOPAb8P+DXbS+UdC7w7lL324Dptt9Y2t67kZOKGINcWUXsWI4F/u/wMJ3t9cA7gL8rx79O9Tqe0dxte52rl56uAGa0yfMwcLCkL0k6Hni6TZ6ICZFgFTF5PUf5b1zSLlSrVw/bXNveQptRFttPAW8B7gA+Dny1qY5GjCbBKmLH8m3gA5L2AyjDgP9M9YZ/gDnAd8r2WuBtZfsk4CVjqP+nVEulI2kasIvtf6AaejyiC/2P2Ca5ZxWxA7G9WtLngDslbQG+B/wB8DVJ5wH/AfxeyX4p8E1J9wE3Af85hiYuAW6S9BjwR6Xe4f+pPb97ZxIxPpm6HhERfS/DgBER0fcSrCIiou8lWEVERN9LsIqIiL6XYBUREX0vwSoiIvpeglVERPS9/w/VweDc3OjBfgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "age_frequency = age_grouped.value_counts(sort=False)\n",
    "age_frequency.rename(\"counts\", inplace=True)\n",
    "ax = age_frequency.plot(kind=\"barh\")\n",
    "ax.set_xlabel(\"counts\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "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>observed_effect</th>\n",
       "      <th>counts</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>age</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>(24.999, 30.0]</th>\n",
       "      <td>2.263298</td>\n",
       "      <td>276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(30.0, 35.0]</th>\n",
       "      <td>3.100541</td>\n",
       "      <td>214</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(35.0, 40.0]</th>\n",
       "      <td>2.433750</td>\n",
       "      <td>177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(40.0, 45.0]</th>\n",
       "      <td>3.801950</td>\n",
       "      <td>197</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(45.0, 50.0]</th>\n",
       "      <td>3.603730</td>\n",
       "      <td>234</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(50.0, 55.0]</th>\n",
       "      <td>5.238368</td>\n",
       "      <td>186</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(55.0, 60.0]</th>\n",
       "      <td>2.871668</td>\n",
       "      <td>132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(60.0, 65.0]</th>\n",
       "      <td>1.402340</td>\n",
       "      <td>85</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(65.0, 70.0]</th>\n",
       "      <td>3.602345</td>\n",
       "      <td>53</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(70.0, 75.0]</th>\n",
       "      <td>-9.985035</td>\n",
       "      <td>12</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                observed_effect  counts\n",
       "age                                    \n",
       "(24.999, 30.0]         2.263298     276\n",
       "(30.0, 35.0]           3.100541     214\n",
       "(35.0, 40.0]           2.433750     177\n",
       "(40.0, 45.0]           3.801950     197\n",
       "(45.0, 50.0]           3.603730     234\n",
       "(50.0, 55.0]           5.238368     186\n",
       "(55.0, 60.0]           2.871668     132\n",
       "(60.0, 65.0]           1.402340      85\n",
       "(65.0, 70.0]           3.602345      53\n",
       "(70.0, 75.0]          -9.985035      12"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "by_age = observed_diff_by_age.to_frame().join(age_frequency)\n",
    "by_age"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In fact, for the largest groups, ages 40 to 55, the observed effect is larger than the overall average observed effect (2.5kg).  \n",
    "This is somewhat similar to [Simpson's paradox](https://en.wikipedia.org/wiki/Simpson%27s_paradox): \n",
    "How come combining groups with a large effect results in a small effect?  \n",
    "\n",
    "To check this, we'll dig even deeper into the distribution of treated vs. control in each of these age groups:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>observed_effect</th>\n",
       "      <th>counts</th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>propensity</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>age</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>(24.999, 30.0]</th>\n",
       "      <td>2.263298</td>\n",
       "      <td>276</td>\n",
       "      <td>228</td>\n",
       "      <td>48</td>\n",
       "      <td>0.173913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(30.0, 35.0]</th>\n",
       "      <td>3.100541</td>\n",
       "      <td>214</td>\n",
       "      <td>160</td>\n",
       "      <td>54</td>\n",
       "      <td>0.252336</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(35.0, 40.0]</th>\n",
       "      <td>2.433750</td>\n",
       "      <td>177</td>\n",
       "      <td>136</td>\n",
       "      <td>41</td>\n",
       "      <td>0.231638</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(40.0, 45.0]</th>\n",
       "      <td>3.801950</td>\n",
       "      <td>197</td>\n",
       "      <td>146</td>\n",
       "      <td>51</td>\n",
       "      <td>0.258883</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(45.0, 50.0]</th>\n",
       "      <td>3.603730</td>\n",
       "      <td>234</td>\n",
       "      <td>181</td>\n",
       "      <td>53</td>\n",
       "      <td>0.226496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(50.0, 55.0]</th>\n",
       "      <td>5.238368</td>\n",
       "      <td>186</td>\n",
       "      <td>133</td>\n",
       "      <td>53</td>\n",
       "      <td>0.284946</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(55.0, 60.0]</th>\n",
       "      <td>2.871668</td>\n",
       "      <td>132</td>\n",
       "      <td>81</td>\n",
       "      <td>51</td>\n",
       "      <td>0.386364</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(60.0, 65.0]</th>\n",
       "      <td>1.402340</td>\n",
       "      <td>85</td>\n",
       "      <td>55</td>\n",
       "      <td>30</td>\n",
       "      <td>0.352941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(65.0, 70.0]</th>\n",
       "      <td>3.602345</td>\n",
       "      <td>53</td>\n",
       "      <td>38</td>\n",
       "      <td>15</td>\n",
       "      <td>0.283019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>(70.0, 75.0]</th>\n",
       "      <td>-9.985035</td>\n",
       "      <td>12</td>\n",
       "      <td>5</td>\n",
       "      <td>7</td>\n",
       "      <td>0.583333</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                observed_effect  counts    0   1  propensity\n",
       "age                                                         \n",
       "(24.999, 30.0]         2.263298     276  228  48    0.173913\n",
       "(30.0, 35.0]           3.100541     214  160  54    0.252336\n",
       "(35.0, 40.0]           2.433750     177  136  41    0.231638\n",
       "(40.0, 45.0]           3.801950     197  146  51    0.258883\n",
       "(45.0, 50.0]           3.603730     234  181  53    0.226496\n",
       "(50.0, 55.0]           5.238368     186  133  53    0.284946\n",
       "(55.0, 60.0]           2.871668     132   81  51    0.386364\n",
       "(60.0, 65.0]           1.402340      85   55  30    0.352941\n",
       "(65.0, 70.0]           3.602345      53   38  15    0.283019\n",
       "(70.0, 75.0]          -9.985035      12    5   7    0.583333"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tx_distribution_by_age = data.a.groupby(age_grouped).value_counts()\n",
    "tx_distribution_by_age = tx_distribution_by_age.unstack(\"qsmk\")\n",
    "tx_distribution_by_age[\"propensity\"] = tx_distribution_by_age[1] / tx_distribution_by_age.sum(axis=\"columns\")\n",
    "\n",
    "by_age = by_age.join(tx_distribution_by_age)\n",
    "by_age"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see that each age group has different proportions between the treated and control, \n",
    "meaning, _individuals in each treatment group have different propensity to quit smoking_.\n",
    "\n",
    "We can go on and visualize the observed effect as a function of the propensity of each age group (blue dots), compared to the overall average (green line) and the effect derived from the causal analysis (orange line):"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = sns.scatterplot(x=\"propensity\", y=\"observed_effect\", data=by_age.reset_index(),\n",
    "                     size=\"counts\", size_norm=(20, 200))\n",
    "ax.axhline(y=causal_effect, linestyle=\"--\", color=\"C1\", label=\"causal effect\", zorder=0)\n",
    "ax.axhline(y=observed_effect, linestyle=\":\", color=\"C2\", label=\"observed effect\", zorder=0)\n",
    "ax.legend();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It can be seen that the ratios of those who quit smoking is smaller (small propensity) in the groups for which the effect is large (the top left of the plot). The groups with small effect contribute disproportionately to the observed effect.\n",
    "The overall observed effect is therefore smaller than what would have been observed if the ratios were the same across all groups. \n",
    "In fact, it would be closer to the 3.5 value that is predicted by the causal analysis.\n",
    "\n",
    "More generally, both the effect size and the prevalence of treatment changes between age groups, and age is therefore a confounder of the effect. \n",
    "Therefore, the overall population effect can only be calculated by weighting the observed group-differences by the size of each age group.  \n",
    "So if we were to average the observed effect and factoring in the size of each age group we would get:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.1002018883702944"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "\n",
    "np.average(by_age[\"observed_effect\"], weights=by_age[\"counts\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Observed effect of 3.1 is closer to the causal effect of 3.5, than to the simple observed effect of 2.5 we originally got.  \n",
    "What we did was to take the already averaged-within-each-age-group effect and average it once more using the sizes of each group.  \n",
    "This is equivalent to providing each individual with a propensity value based only on their age group, and then applying an inverse-probability-weighting scheme to obtain a slightly-more corrected effect.  \n",
    "Let's do IPW by hand:\n",
    "#### Building an age-based IPW model by hand"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "ename": "TypeError",
     "evalue": "Object with dtype category cannot perform the numpy op multiply",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-28-916d9792bddf>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;31m# i.e. the treated get their propensity (=probability to be treated)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0;31m#      and the control get 1 minus that propensity (=probability to by untreated)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m \u001b[0mindividual_age_weights\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0ma\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0mindividual_age_weights\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0ma\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mindividual_age_weights\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[0;31m# We then inverse these propensity:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/causal/lib/python3.8/site-packages/pandas/core/ops/common.py\u001b[0m in \u001b[0;36mnew_method\u001b[0;34m(self, other)\u001b[0m\n\u001b[1;32m     67\u001b[0m         \u001b[0mother\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mitem_from_zerodim\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mother\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     68\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 69\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mmethod\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mother\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     70\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     71\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mnew_method\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/causal/lib/python3.8/site-packages/pandas/core/arraylike.py\u001b[0m in \u001b[0;36m__mul__\u001b[0;34m(self, other)\u001b[0m\n\u001b[1;32m    106\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0munpack_zerodim_and_defer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"__mul__\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    107\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m__mul__\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mother\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 108\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_arith_method\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mother\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moperator\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmul\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    109\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    110\u001b[0m     \u001b[0;34m@\u001b[0m\u001b[0munpack_zerodim_and_defer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"__rmul__\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/causal/lib/python3.8/site-packages/pandas/core/series.py\u001b[0m in \u001b[0;36m_arith_method\u001b[0;34m(self, other, op)\u001b[0m\n\u001b[1;32m   5524\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   5525\u001b[0m         \u001b[0;32mwith\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0merrstate\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mall\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m\"ignore\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 5526\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mops\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marithmetic_op\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlvalues\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrvalues\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mop\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   5527\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   5528\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_construct_result\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mresult\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mname\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mres_name\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/causal/lib/python3.8/site-packages/pandas/core/ops/array_ops.py\u001b[0m in \u001b[0;36marithmetic_op\u001b[0;34m(left, right, op)\u001b[0m\n\u001b[1;32m    216\u001b[0m         \u001b[0;31m# Timedelta/Timestamp and other custom scalars are included in the check\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    217\u001b[0m         \u001b[0;31m# because numexpr will fail on it, see GH#31457\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 218\u001b[0;31m         \u001b[0mres_values\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mop\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mleft\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mright\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    219\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    220\u001b[0m         \u001b[0;31m# TODO we should handle EAs consistently and move this check before the if/else\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/causal/lib/python3.8/site-packages/pandas/core/arrays/categorical.py\u001b[0m in \u001b[0;36m__array_ufunc__\u001b[0;34m(self, ufunc, method, *inputs, **kwargs)\u001b[0m\n\u001b[1;32m   1466\u001b[0m         \u001b[0;31m# for all other cases, raise for now (similarly as what happens in\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1467\u001b[0m         \u001b[0;31m# Series.__array_prepare__)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1468\u001b[0;31m         raise TypeError(\n\u001b[0m\u001b[1;32m   1469\u001b[0m             \u001b[0;34mf\"Object with dtype {self.dtype} cannot perform \"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1470\u001b[0m             \u001b[0;34mf\"the numpy op {ufunc.__name__}\"\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mTypeError\u001b[0m: Object with dtype category cannot perform the numpy op multiply"
     ]
    }
   ],
   "source": [
    "# Give each individual their propensity according their age group:\n",
    "individual_age_weights = age_grouped.map(by_age[\"propensity\"])\n",
    "\n",
    "# Give each individual the pobability to be in their original treatment group,\n",
    "# i.e. the treated get their propensity (=probability to be treated) \n",
    "#      and the control get 1 minus that propensity (=probability to by untreated)\n",
    "individual_age_weights = (data.a * individual_age_weights) + ((1-data.a) * (1-individual_age_weights))\n",
    "\n",
    "# We then inverse these propensity:\n",
    "individual_age_weights = 1 / individual_age_weights\n",
    "\n",
    "# We use the inversed propensities to create a weighted outcome:\n",
    "weighted_outcome = individual_age_weights * data.y\n",
    "\n",
    "# Finaly, we take the weighted average of the weighted outcome in each treatment group:\n",
    "weighted_outcome = weighted_outcome.groupby(data.a).sum() / individual_age_weights.groupby(data.a).sum()\n",
    "\n",
    "# And we can \n",
    "weighted_outcome[1] - weighted_outcome[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Indeed we got the same effect as the weighted-average-of-averages above.\n",
    "\n",
    "However, remember we had only 10 unique propensity values, because we arbitrarily binned the age variable into 10 age groups.  \n",
    "Meanwhile, using a machine learning model allows us to model the propensity using `age` as a continuous variable.  \n",
    "Let's examine what happens when using the age continuously, by building an IPW model that uses `age` as its sole feature when modeling propensity to treat."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3.098787136071221"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learner = LogisticRegression(penalty='none', solver='lbfgs', max_iter=500)\n",
    "ipw = IPW(learner)\n",
    "ipw.fit(data.X[[\"age\"]], data.a)\n",
    "outcomes = ipw.estimate_population_outcome(data.X[[\"age\"]], data.a, data.y)\n",
    "outcomes[1] - outcomes[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Indeed we got an almost identical effect, correcting only for age in our IPW model.  \n",
    "We can only assume that when correcting for more variables and eliminating more spuriuous correlations in the process, we will get to the causal effect of 3.5 kg."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary\n",
    "1. We saw that there is a discrepancy between the \"effect\" using simple descriptive statistics, and the causal effect resulting from a causal analysis.\n",
    "2. The discrepancy originated from the groups (quitting or not quitting smoking) having different distributions of charachteristics, and thus, are not comparable.\n",
    "3. We discovered that participants' age had the largest distributional shift among the treatment groups and focused on it.\n",
    "4. We saw that the observed effect differed across different age groups, and that different age groups had different number of people, suggesting a simple average mixes different effects indifferently. \n",
    "5. We demonstrated that when accounting for the number of participants in each age group, the \"effect\" shifted from the original observed-effect towards the causal-effect.\n",
    "7. We saw we get similar result when building a causal model that controls only for age (and not for all confounders)."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.10 ('causal')",
   "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": "3.8.10"
  },
  "vscode": {
   "interpreter": {
    "hash": "69f88c614e1e3c90b1e7d1bcdbafb917f9042edfca4477b05464240f7cf41bf5"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}