{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Policy Learning I - Binary Treatment\n",
    "\n",
    "\n",
    "\n",
    "<font color=\"#fa8072\">Note: this chapter is in 'beta' version and may be edited in the near future.</font>\n",
    "\n",
    "A few chapters ago, we learned how to estimate the average effect of a binary treatment (ATE), that is, the value of treating everyone in a population versus treating no one. Once that was established, we asked whether certain subgroups could react differently to the treatment, as we learned how to estimate such heterogeneous treatment effects (HTE). Then, in the previous chapter, we learned how to aggregate these heterogeneous effects to estimate the average outcome that would be attained if treatment assignment were to follow a particular rule, that is, if we were to treat only individuals with certain observable characteristics (policy evaluation). In this chapter, we will learn how to search the space of available treatment rules to approximately _maximize_ the average outcome across the population. That is, we will answer questions of the type: \"_who_ should be treated?\" We'll call this problem **policy learning**. \n",
    "\n",
    "We'll make a distinction between parametric and non-parametric policies, just as we did with predictive models. Parametric policies are simpler and depend only on a fixed number of parameters, whereas nonparametric policies can increase in complexity with the data. As we'll discuss below, there will be situations in which one or the other will be more appropriate.\n",
    "\n",
    "For now, we'll work with the same toy simulation setting that we used in the previous chapter. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import statsmodels.formula.api as smf\n",
    "import statsmodels.api as sm\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib\n",
    "import matplotlib.mlab as mlab\n",
    "import random \n",
    "import econml\n",
    "import time\n",
    "from patsy import dmatrices\n",
    "from sklearn.linear_model import Lasso, LassoCV\n",
    "from sklearn.model_selection import train_test_split\n",
    "from scipy.stats import norm, expon, binom\n",
    "from econml.grf import RegressionForest, CausalIVForest as instrumental_forest\n",
    "from econml.dml import CausalForestDML as causal_forest\n",
    "from econml.policy import PolicyForest, PolicyTree \n",
    "random.seed(12)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>x_1</th>\n",
       "      <th>x_2</th>\n",
       "      <th>x_3</th>\n",
       "      <th>x_4</th>\n",
       "      <th>y</th>\n",
       "      <th>w</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.617824</td>\n",
       "      <td>0.722115</td>\n",
       "      <td>0.313114</td>\n",
       "      <td>0.340408</td>\n",
       "      <td>0.384143</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.831421</td>\n",
       "      <td>0.221606</td>\n",
       "      <td>0.660022</td>\n",
       "      <td>0.772754</td>\n",
       "      <td>-0.033141</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.246041</td>\n",
       "      <td>0.338020</td>\n",
       "      <td>0.557677</td>\n",
       "      <td>0.595021</td>\n",
       "      <td>-0.269752</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.675558</td>\n",
       "      <td>0.025890</td>\n",
       "      <td>0.920674</td>\n",
       "      <td>0.702211</td>\n",
       "      <td>-0.095722</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        x_1       x_2       x_3       x_4         y  w\n",
       "0  0.617824  0.722115  0.313114  0.340408  0.384143  1\n",
       "1  0.831421  0.221606  0.660022  0.772754 -0.033141  1\n",
       "2  0.246041  0.338020  0.557677  0.595021 -0.269752  0\n",
       "3  0.675558  0.025890  0.920674  0.702211 -0.095722  0"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# A randomized setting.\n",
    "n, p, e = 1000, 4, .5\n",
    "x = np.reshape(np.random.uniform(0, 1, n*p), (n, p))\n",
    "w = binom.rvs(1, p = e, size = n)\n",
    "# w = np.random.binomial(1, e, n)\n",
    "# x = np.random.normal(size = (n, p))\n",
    "y = e * (x[:, 0] - e) + w * (x[:, 1] - e) + .1 * np.random.normal(0, 1, n)\n",
    "\n",
    "data = pd.DataFrame(x)\n",
    "data.columns = ['x_1', 'x_2', 'x_3', 'x_4']\n",
    "data[\"y\"], data[\"w\"] = y, w\n",
    "\n",
    "# data = pd.read_csv(\"cap6_data/test.csv\")\n",
    "\n",
    "outcome, treatment, covariates = \"y\", \"w\", list(data.columns)[1:5]\n",
    "data.head(4)\n",
    "\n",
    "# covariates"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Non-parametric policies\n",
    "\n",
    "In the HTE chapter we define the conditional average treatment effect (CATE) function\n",
    "\n",
    "$$\n",
    "  \\tau(x) := \\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0) | X_i = x],\n",
    "$$ (cate)\n",
    "\n",
    "that is, the average effect of a binary treatment conditional on observable charateristics. If we knew {eq}`cate`, then a natural policy would be to assigns individuals to treatment their CATE is positive,\n",
    "\n",
    "$$\n",
    "  \\pi^{*} = \\mathbb{I}\\{\\tau(x) \\geq 0\\}.\n",
    "$$\n",
    "\n",
    "More generally, if treating that individual costs a known amount $c(x)$, \n",
    "\n",
    "$$\n",
    "  \\pi^{*} = \\mathbb{I}\\{\\tau(x) \\geq c(x)\\}.\n",
    "$$\n",
    "\n",
    "Of course, we don't know {eq}`cate`. However, we can obtain an estimate $\\widehat{\\tau}(\\cdot)$ using any flexible (i.e., non-parametric) method as we learned in the HTE chapter, and then obtain a policy estimate\n",
    "\n",
    "$$\n",
    "  \\hat{\\pi}(x) = \\mathbb{I}\\{ \\widehat{\\tau}(x) \\geq 0\\},\n",
    "$$\n",
    "\n",
    "replacing the zero threshold by some appropriate cost function if needed.\n",
    "\n",
    "Once we have an estimated policy, we need to estimate its value. To obtain accurate estimates, we must ensure appropriate **data-splitting**. We cannot estimate and evaluate a policy using the same data set, because that would lead to an overestimate of the value of the policy. One option here is to divide the data into training and test subsets, fit $\\widehat{\\tau}(\\cdot)$ in the training subset and evaluate it in the test subset. This is analogous to what we saw in prediction problems: if we try to evaluate our predictions on the training set, we will overestimate how good our predictions are. \n",
    "\n",
    "The next snippet estimates the conditional treatment effect function via a Lasso model with splines. Note the data splitting.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'y ~  bs(x_2 , df = 5) * w + bs(x_3 , df = 5) * w + bs(x_4 , df = 5) * w + bs(y , df = 5) * w '"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Preparing to run a regression with splines (\\\\piecewise polynomials).\n",
    "# Note that if we have a lot of data we should increase the argument `df` below.\n",
    "# The optimal value of `df` can be found by cross-validation\n",
    "# i.e., check if the value of the policy, estimated below, increases or decreases as `df` varies. \n",
    "\n",
    "def bs_x(x_n, d_f = 5, add = True):\n",
    "    bs_n = \" bs(\" + covariates[x_n] + \" , df = \" + str(d_f) + \") * w \"\n",
    "    if add:\n",
    "        bs_n = \"+\" + bs_n\n",
    "    return bs_n\n",
    "fmla_xw = \"y ~ \" +  bs_x(0, add = False) + bs_x(1) + bs_x(2) + bs_x(3)\n",
    "fmla_xw"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Data-splitting\n",
    "## Define training and evaluation sets\n",
    "\n",
    "data_train, data_test = train_test_split(data, test_size = .5, random_state=1) \n",
    "\n",
    "## Contruct matrices \n",
    "\n",
    "y_train, xw_train = dmatrices(fmla_xw, data_train)\n",
    "y_test, xw_test = dmatrices(fmla_xw, data_test)\n",
    "\n",
    "y_test, y_train = np.ravel(y_test), np.ravel(y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.086e-02, tolerance: 3.938e-03\n"
     ]
    }
   ],
   "source": [
    "# Fitting the outcome model on the *training* data\n",
    "model_m = LassoCV(cv = 10, random_state=12)\n",
    "model_m.fit(xw_train, y_train)\n",
    "data_0 = data_test.copy()\n",
    "data_1 = data_test.copy()\n",
    "\n",
    "# Predict outcome E[Y|X,W=w] for w in {0, 1} on the *test* data\n",
    "data_0[treatment] = 0\n",
    "data_1[treatment] = 1\n",
    "\n",
    "## Construct matirces\n",
    "y0, xw0 = dmatrices(fmla_xw, data_0)\n",
    "y1, xw1 = dmatrices(fmla_xw, data_1)\n",
    "\n",
    "# Predict values\n",
    "mu_hat_1 = model_m.predict(xw1)\n",
    "mu_hat_0 = model_m.predict(xw0)\n",
    "\n",
    "# Extract rows \n",
    "n_row, n_col = data_test.shape\n",
    "\n",
    "# Computing the CATE estimate tau_hat \n",
    "tau_hat = mu_hat_1 - mu_hat_0\n",
    "\n",
    "# Assignment if tau.hat is positive (or replace by non-zero cost if applicable)\n",
    "pi_hat = tau_hat > 0\n",
    "\n",
    "# Estimate assignment probs e(x). \n",
    "# (This will be useful for evaluation via AIPW scores a little later)\n",
    "\n",
    "# In randomized settings assignment probabilities are fixed and known.\n",
    "e_hat = np.repeat(.5, n_row)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "On the test set, we can evaluate this policy as we learned in the previous chapter. In randomized settings, a simple estimator based on the difference in means is available.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0887461287957061, \\  0.0164705393732257\\right)$"
      ],
      "text/plain": [
       "(0.08874612879570612, 0.016470539373225738)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Only valid in randomized settings.\n",
    "\n",
    "a = pi_hat == 1\n",
    "Y = data_test[outcome]\n",
    "W = data_test[treatment]\n",
    "\n",
    "## Extract, value estimate and standard error\n",
    "def extr_val_sd(y_eval, w_eval, a, cost = 0):\n",
    "    c_1 = a & (w_eval == 1)\n",
    "    c_0 = np.logical_not(a) & (w_eval == 0)\n",
    "\n",
    "    y_eval_main = pd.DataFrame(y_eval)\n",
    "    y_eval_main['c_1'] = c_1\n",
    "    y_eval_main['c_0'] = c_0\n",
    "\n",
    "    y_0 = y_eval_main.loc[y_eval_main['c_0'] == True]['y']\n",
    "    y_1 = y_eval_main.loc[y_eval_main['c_1'] == True]['y']\n",
    "\n",
    "    mean_1 = np.mean(y_1 - cost) * np.mean(a)\n",
    "    mean_0 = np.mean(y_0) * np.mean(np.logical_not(a))\n",
    "    val_est = mean_1 + mean_0\n",
    "\n",
    "    var_1 = np.var(y_eval[c_1]) / np.sum(c_1) * np.mean(a)**2\n",
    "    var_0 = np.var(y_eval[c_0]) / np.sum(c_0) * np.mean(np.logical_not(a))**2\n",
    "    var_sqr = np.sqrt(var_1 + var_0)\n",
    "    return val_est, var_sqr\n",
    "\n",
    "extr_val_sd(Y, W, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In randomized settings and observational settings with unconfoundedness, an estimator of the policy value based on AIPW scores is available. In large samples, it should have smaller variance than the one based on sample averages.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd0AAAAVCAYAAAAD8TQDAAAL00lEQVR4nO2dfZRXRRnHP9AiIipbKm4vIrAFbYoKJmmJQiGWlAcry9OJ7A3ymBEW+ILHWOwoWEFkr/SmRJysBA2Pihh6Qqy0FE5yBMFgQV6kgEAMMF3pj+9Mv8v9zf397sy9v112u99z9tzde2fmeWaemWdmnueZ2S7Nzc0UKFCgQIECBWqPrgnv5wL/AHq2IS8FChQoUKBAZ8CZwEHg8/EPrkn3LGAsMAP4d+zbW4CfA1uBl4EWYDbw+gCmfMv6KPBd4FHgRVShX1ahcSuwFHge2A/sAlYAU4HjEvJ0AcYBjwMvoTb4K3AF7vYKoRFSF588nzbfK/20VqEH8MlI+rLOk6EukF7+xxnadwPPoTbeAywHPkfywtGHhoWv7MGv/lnqkgc60vgN7VcA70Nt/ILhbSvwIHCRI62PzEPl10LyOHwhAz9Z8vjwZBGi60YDS4DNJs964LfAOTnRyGNMpdFzFmn71pPAPcDXgaOjH7o4zMtLgKHAG00FLBqBPwK9gd8Ba0y6EcCzwHuAnVUYzlLWSuB01Kk2A28H5qMGS8J/gKeAZyjt3M8G3oka62wk3CjmA58w6RcB+4ALgCZgHvCpHGiE1MUnzxnAmIRyhgHvBe4DPliB3knA08DrUKcZB/w0I18WPvK/AvghsA14BNgEnAh8GOgFLAAuRQMmlIaFr+x96x9alzzQ0cavb3qLbwCTTZ4HgB3ACWjn8Xvgmlh6H5mHyq8FqEeLkjheAr4VyE+WPD48WfjqultRe+9EE9AO4K3AxUCd4Su+kPKlkXVMpdVz4N+3hqKF0A3ALfZlXSzRAGCkIbo/9u0HaJBNQCtQi1nA1cDNqAHSIKSsq1FlnwPORw1cDccCBxzvbwamANcDV0beX4I67wbUYDvM+yOQ8MaizrMwA43QuvjkWWl+XPiTef64Qv4uwO1osCwEJuXEl4WP/NeiQXof8Fok7RTgCeAjaIAtyEADwmQPfvUPrUse6GjjN6RfjUNKcS4wHinwKLrF/vaVeRb57Qaaq/Af0gdD+21anqLw0XUNSG9sB05DE6jFCOBh4CbKJ11ffZpFJj56zrdvYeivAb6ALMevQfnW+7OGkV/H3jcCo9Dq6Puxb1OROWMs6XzAoWU9AqzDbxfgEh7Ab8zzbbH3l5jnTEqdF9TAN5rfr8pIA8LqEpInjkFopbgFddIkTEC74c9Q7mLIypev/B8G7uXQAQUy7/zI/D48Iw0Ikz341T+kLnmgI45f3/TdkWLehFspArwS+9tX5rWWX0gfDO23IfDRdSej+eVxDp1wQbLdi3aJWWhANpmk1XMhfcviTqAPsjwA5ZPuSOTr+3Ps/QjzXEJ55fYCjwFHIYVeDXmWFYoPmeffYu8bzHO9I499NwytIkNptCfGm+fPSPbpNqFV2XeAZTXgIU/5247+ag408pR9CJLqkgc62/h14QKkxBcavkYD1wJfJtl/mKfMq8mvOzKNTzE8jUAmzaz8ZKlDGp7SwKXr1qHJaShwfCz9ecAxyCSbhUY1VJKJj54L6VsWj0XKAA41L/dEvsDVlM/6A81zbULB69DqdwBygldCnmWlxSRkr++FfAPnIuHNiKWzK8V+jjL6m2ed+X1NII32Qg80wFpJ9lnUIR/QJjQQa4G85G99QgCLc6CRRfZZUakueaCjj980OMs8D6DAm1Nj35eh4Kx/Rt7lJfM08mtAYyuKDWiX9YcM/GSpQxqeXEij63ahiWkW8s/eg8y4jcgc/BAyu2ahUQmVZOKr50L6lsVfzPM8+yK6030zWuVsc2TsZZ57Epiy7+sTvteqrLSYhExfE5HwFiPFEG8ka3L9CvCGyPtuwLTI364IzbQ02gsfQ226mPLALouvAYNR9HPcp58X8pL/DNT570fRg1lpZJF9VlSqSx7o6OM3DXqb52Rkkh6GdlOnoV35eShqNoq8ZF5NfrejqNcGtLkZBMwB+qKAnNMz8BNah7Q8uZBW181G/tQ65BO9DgU1PQ/cQbnZOYRGEirJxFfPhfQtiz1osu5jX0QnXRuO/a8UTHQ0NCBfdQPqBP3RimVILN2dSECNaHU2B5kfVqKG3mTSxc1qPjTaC9a0PCfh+7vQqm8mpWCrwxUTgK+ilfvYnMrMIvssqEVd/h9hddmraCe1HEXhPo38nptRQFbUHJiHzNPIbxryPW5HkcWrUKDZLGSBas7AT2gd0vLkQlpddw1wF5pgG9HkfiYye89H0cBZabhQSSYhei6kb0Wxi4iJPTrp2hn/SEcmu3rt5fgWfb87me+alOWL7eiM1Si0yPhF7Hsr8h1ch1ZUl5ufdcC7kc8KKq/QqtFoD5yC+N+MVn5x1CE+11IKvqgVssr/KqRUnkE+qF050chD9r5IU5c80FnGbyVYeitQkFcU+yjtdoZG3meVeVb52UAfa3oM4SfvfhvnqRIq6brh6MjQIrQLX4/k8BSaqLagibE/leGrTyvJJFTP7TZPn74VRQ8iO+ropGuF4jqA/Kx5Dkgo1EaUJfl5alVWKDYioZxCuZP/FdRZBqEFSD0689pieNuB/B5ZaLQ1qgVQHY3k0YRMIQcjP1NNmp+Yv2dn5CWL/CeiIyqr0IBKOsQfSiMv2afBRNLVJQ90tvHrguVrd8J3a8HrEXsfKvOJZJefNZVGI71D+Mmz37p4qgaXrrN3ALiOeu1Dx2m6IjNvKI04JlJZJqF6LrRvgepYT2TREw2k2oYafCDlsA03yhQSNVUcgw7D76M86tmFPMvKgjeZZ5qbmQAuQxGAv6ohjVrgSGRiaUWTrgsvV/g2BA2M5ajzZTU9h8r/WuSnWYkiAXeQjLz7WIjsK8GnLnmgM47fOJYiZfkOB19QCn5JO/lUknle8rMR3q7IYx9+8szjw1MUcV3X3Txdx4Ki713Hb9LSiCKNTEL1XJa+NRCZyVfaF9Gd7kEUhXU8ujUkir8jh3Ff4Iuxb9PQqmge5VHPjegmmejB4dCyfDEAtwmsKzpz1RvdqhP3YR/ryHMG8E2TNho9F0qjLXEpCqJ4gOQAqv3o+jPXzyKTZq75O36G2xch8r8RtfuTKPijmpIL7WM+sg+Fb11APrGDKPAjBCHt4Rq7oWW1BTai85p90FGOKEYBF6KdSjyS1VfmvvJrwr1r7At8z/wevSAipA/65vHlCfx13aPmOR4F6UbxAbQ4O2DyhNKwSCuTUD0X2regtIj5344/fiPVAnSDx4XoJpgorkQVvs1UbDVySo9ApqQbHASXokPS/TjUFh5S1hhKVxvas2nnIIUEaujojSIXAdPRymUDClc/ETm8+yPTwzgHnYeQcFYhf0gTOpe1H/lOtuZAw7cuoXmgZFqudANVFoTw5SP/y9HNNa1oIE9w8NASoRdCw8JH9hZjSF//0LpEAzlC4dseSWM3pCzw7ye+6UGLgMEoGGg08sH1M+W0ImUaj7r2kXmI/D6OfJfLkPLeixY0o5EV6n4OvXIxpA/65vHlCfx13V3oHO5I1D/uNmmakOm5C/JD78xAA8LHlC9C+hZoUm5F16UC5XcvH4F2Qy1oEMVxEqrg+5HvdxtqzGm4d3MtJA9c37KaKdndXdiIVmoWp6JovHPR5ez1aPW9FoXZ34Y78GEyMss0Ihv9FrRLnI4CkaIIpeFbl9A8TcgPstl8CzFzW7pJd5KG8AXp5V+tfNCZwuEZaFj4yD4tf9H6V0sL7rqsMDydnMB3Wvi0RwvJY9e3LPDvJ77pLU5AR0IuRvfHv4iU8XTkR4zDR+bVeIJy+Z2PdMRgSsdzdiNz4zzzczCQn9A8vjxBmK7rhiary5Bp9iiT5gmTfkkONJoJ1w+ucpL0HPj3rV5oofAgkbvwXf/w4Hp0OfMQNNgLFCjQfqhHK/6ZlF+oXqBAgcMXX0ILhWFo9w64/+3Rt9HZrpvahq8CBQpUwDAUnTqrvRkpUKBAavRAG9gFRCZcKPfpgpzbY5F/pidtHxBRoECBEu7FfXa+QIEChy/6ojiaO+IfXJMuyMG+rHb8FChQoECBAp0Wq0m41ctlXi5QoECBAgUK1AD/BdYA2tbOlGUzAAAAAElFTkSuQmCC\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0139311741020912, \\  0.0116925995383246\\right)$"
      ],
      "text/plain": [
       "(0.013931174102091154, 0.01169259953832457)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Valid in randomized settings and observational settings with unconfoundedness and overlap.\n",
    "Y = data_test[outcome]\n",
    "W = data_test[treatment]\n",
    "\n",
    "# AIPW \n",
    "gamma_hat_1 = mu_hat_1 + W / e_hat * (Y - mu_hat_1)\n",
    "gamma_hat_0 = mu_hat_0 + (1 - W) / (1 - e_hat) * (Y - mu_hat_0)\n",
    "gamma_hat_pi = pi_hat * gamma_hat_1 + (1 - pi_hat) * gamma_hat_0\n",
    "\n",
    "## Print the value_estiamte and standard error\n",
    "ve = np.mean(gamma_hat_pi)\n",
    "std =  np.std(gamma_hat_pi) / np.sqrt(len(gamma_hat_pi))\n",
    "\n",
    "ve, std"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Above we used a flexible linear model, but in fact we can also use any other non-parametric method. The next example uses `econml.grf`. An advantage of using `econ.grf` is that we can leverage [out-of-bag predictions](https://github.com/grf-labs/grf/blob/master/REFERENCE.md#out-of-bag-prediction), so explicit data splitting is not necessary."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make a causal_forest object\n",
    "forest = causal_forest(\n",
    "    model_t = RegressionForest(),\n",
    "    model_y = RegressionForest(),\n",
    "    n_estimators = 200, min_samples_leaf = 5,\n",
    "    max_depth = 50, verbose = 0, \n",
    "    random_state = 2\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using the entire data\n",
    "x = data[covariates]\n",
    "y = data[outcome]\n",
    "w = data[treatment]\n",
    "\n",
    "# Tune and Fit \n",
    "forest.tune(y, w, X=x, W=None)\n",
    "forest_oob = forest.fit(y, w, X=x, W=None)\n",
    "\n",
    "# Extract residuals\n",
    "residuals = forest.fit(y, w, X = x, W = None, cache_values=True).residuals_\n",
    "\n",
    "# Get \"out-of-bag\" predictions\n",
    "tau_hat_oob = forest_oob.effect(x)\n",
    "pi_hat = tau_hat_oob > 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, to evaluate the value of this policy in a randomized setting, we can use the following estimator based on sample averages.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0695399567615397, \\  0.00835655203113492\\right)$"
      ],
      "text/plain": [
       "(0.06953995676153975, 0.008356552031134916)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Only valid in randomized settings.\n",
    "# We can use the entire data because predictions are out-of-bag\n",
    "a = pi_hat == 1\n",
    "\n",
    "# Using a Extract function, return, value estimate and standard error\n",
    "extr_val_sd(y, w, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And here's how to produce an AIPW-based estimate. Note that that estimates of the propensity scores (`w_hat`) and outcome model (`mu_hat_1`, `mu_hat_0`) are also [out-of-bag](https://github.com/grf-labs/grf/blob/master/REFERENCE.md#out-of-bag-prediction), ensuring appropriate sample splitting.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0204158668608208, \\  0.00794957243031285\\right)$"
      ],
      "text/plain": [
       "(0.020415866860820784, 0.007949572430312846)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Valid in randomized settings and observational settings with unconfoundedness and overlap.\n",
    "tau_hat = forest_oob.effect(x)\n",
    "\n",
    "# Retrieve relevant quantities.\n",
    "e_hat = w - residuals[1] # P[W=1|X]\n",
    "y_hat = y - residuals[0] \n",
    "\n",
    "mu_hat_1 = y_hat + (1 - e_hat) * tau_hat # E[Y|X,W=1] = E[Y|X] + (1 - e(X)) * tau(X) \n",
    "mu_hat_0 = y_hat - e_hat * tau_hat # E[Y|X,W=0] = E[Y|X] - e(X) * tau(X)\n",
    "\n",
    "## Compute AIPW score\n",
    "gamma_hat_1 = mu_hat_1  + w / e_hat * (y - mu_hat_1)\n",
    "gamma_hat_0 = mu_hat_0 + (1 - w) / (1 - e_hat) * (Y -mu_hat_0) # T can be W\n",
    "gamma_hat_pi = pi_hat * gamma_hat_1 + (1 - pi_hat) * gamma_hat_0\n",
    "\n",
    "np.mean(gamma_hat_pi), np.std(gamma_hat_pi)/ np.sqrt(len(gamma_hat_pi))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "A technical note. It's easy to get confused and try to \"estimate\" a nonparametric policy using AIPW scores, as in \"$\\hat{\\pi}(X_i) = \\mathbb{I}\\{ \\widehat{\\Gamma}_{i,1} \\geq  \\widehat{\\Gamma}_{i,0} \\}$\". *This is incorrect*. AIPW scores are very noisy and should never be used \"pointwise\" like this. They should be used as part of an average (as above), or some other form of aggregation (as we'll see in the next section).\n",
    "\n",
    "\n",
    "## Parametric policies\n",
    "\n",
    "In many settings, there are good reasons to constrain the policy to belong to a smaller function class $\\pi$. The set $\\pi$ may contain only policies that, for example, are transparent and easy to explain to stakeholders, or that are easily implemented in the field. It may also be the case that the set of available policies $\\pi$ encodes other desirability criteria, such as satisfying certain budget constraints or depending only on a subset of observable characteristics.\n",
    "\n",
    "Estimating such a policy from data is finding an approximate solution to the following constrained maximization problem,\n",
    "\n",
    "$$ \n",
    "  \\pi^{*} = \\arg\\max_{\\pi \\in \\Pi} \\mathop{\\mathrm{E}}[Y(\\pi(X_i))].\n",
    "$$  (param-pi-oracle)\n",
    "\n",
    "\n",
    "Following [Athey and Wager (2020, Econometrica)], we will use the following em\\\\pirical counterpart of {eq}`param-pi-oracle`,\n",
    "\n",
    "$$ \n",
    "  \\hat{\\pi} = \\arg\\min_{\\pi \\in \\Pi} \\frac{1}{n} \\sum_{i=1}^{n} \\widehat{\\Gamma}_{i,\\pi(X_i)}\n",
    "$$ (param-pi-problem)\n",
    "\n",
    "where $\\widehat{\\Gamma}_{i,\\pi(X_i)}$ are AIPW scores as defined in the previous chapter. As reminder, \n",
    "\n",
    "$$ \n",
    "  \\widehat{\\Gamma}_{i,\\pi(X_i)} = \\pi(X_i)\\widehat{\\Gamma}_{i,1} + (1 - \\pi(X_i))\\widehat{\\Gamma}_{i,0},\n",
    "$$ \n",
    "\n",
    "\n",
    "where\n",
    "\n",
    "$$  \n",
    "\\begin{align}\n",
    "    \\widehat{\\Gamma}_{i,1} \n",
    "    &= \\hat{\\mu}^{-i}(X_i, 1) + \\frac{W_i}{\\hat{e}^{-i}(X_i)} \\left(Y_i -\\hat{\\mu}^{-i}(X_i, 1)\\right), \\\\\n",
    "    \\widehat{\\Gamma}_{i,0} \n",
    "    &= \\hat{\\mu}^{-i}(X_i, 0) . \\frac{1-W_i}{1-\\hat{e}^{-i}(X_i)} \\left(Y_i -\\hat{\\mu}^{-i}(X_i, 0)\\right).\n",
    "\\end{align}\n",
    "$$ (aipw)\n",
    "\n",
    "Here we use shallow tree policies as our main example of parametric policies. The `R` package `policytree` to find a policy that solves {eq}`param-pi-problem`. In the example below, we'll construct AIPW scores estimated using `grf`, though we could have used any other non-parametric method (with appropriate sample-splitting). See this short [tutorial](https://grf-labs.github.io/policytree/) for other examples using these two packages.\n",
    "\n",
    "Let's walk through an example for the data simulated above. The first step is to construct AIPW scores {eq}`aipw`. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Randomized setting: pass the known treatment assignment as an argument.\n",
    "forest.tune(y, w, X=x, W=None)\n",
    "forest_prm_p = forest.fit(y, w, X=x, W=None, cache_values = True)\n",
    "\n",
    "## Extract Gamma_hat_0 and gamma_hat_1 => R: double_robust_scores(forest)\n",
    "def double_robust_score(forest_model, y, w, x):\n",
    "    residuals = forest_model.residuals_\n",
    "\n",
    "    tau_hat = forest_model.effect(x)\n",
    "\n",
    "    e_hat = w - residuals[1]\n",
    "    y_hat = y - residuals[0]\n",
    "    mu_hat_1 = y_hat + (1 - e_hat) * tau_hat # E[Y|X,W=1] = E[Y|X] + (1 - e(X)) * tau(X) \n",
    "    mu_hat_0 = y_hat - e_hat * tau_hat # E[Y|X,W=0] = E[Y|X] - e(X) * tau(X)\n",
    "    \n",
    "    gamma_hat_1 = mu_hat_1  + w / e_hat * (y - mu_hat_1)\n",
    "    gamma_hat_0 = mu_hat_0 + (1 - w) / (1 - e_hat) * (y - mu_hat_0) # T can be W\n",
    "    \n",
    "    return gamma_hat_1, gamma_hat_0\n",
    "    \n",
    "gamma_hat_0, gamma_hat_0 = double_robust_score(forest_prm_p, y, w, x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "gamma_mtrx = pd.DataFrame(\n",
    "    {\n",
    "        \"gamma_hat_0\" : gamma_hat_0,\n",
    "        \"gamma_hat_1\": gamma_hat_1\n",
    "    }\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, to ensure appropriate sample splitting, we divide our data into training and test subsets. We estimate the policy on the training subset and estimate its value on the test subset.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set train size\n",
    "train = int(n/2)\n",
    "\n",
    "# Estimate the policy on the training subset \n",
    "policy = PolicyTree(\n",
    "    max_depth = 2, random_state = 2\n",
    ").fit(x.iloc[ : train], gamma_mtrx.iloc[ : train])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predict on the test subsets\n",
    "pi_hat = policy.predict(x.iloc[int(n/2):, ]) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "We can plot the tree.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "plt.figure(figsize=(25, 5))\n",
    "policy.plot(treatment_names = [\"Control\", \"Treatment\"])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note how the treatment rule is rather transparent, in that whether or not each individual is treated depends only on a couple of if-statements. This can be very attractive in settings in which it's important to explain the policy to stakeholders, or reason about its consequences in terms of fairness (e.g., is it okay that these particular subgroups get the treatment?), manipulability (e.g., will individuals lie about their observable characteristics to get a better outcome?), and so on. \n",
    "\n",
    "To evaluate the policy, we again use what we learned in the previous chapter, remembering that we can only use the test set for evaluation. In randomized settings, we can use the following estimator based on sample averages.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0276991701751816, \\  0.011469299931345\\right)$"
      ],
      "text/plain": [
       "(0.027699170175181587, 0.011469299931345028)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "a = pi_hat == 1\n",
    "y1 = data.iloc[train : ][outcome]\n",
    "w1 = data.iloc[train : ][treatment]\n",
    "extr_val_sd(y1, w1, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the remaining AIPW scores produces an estimate that, in large samples, has smaller standard error.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0276621802859057, \\  0.0250615788386545\\right)$"
      ],
      "text/plain": [
       "(0.027662180285905655, 0.025061578838654543)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Using the remaining AIPW scores produces an estimate that, in large samples, has smaller standard error.\n",
    "gamma_hat_pi = pi_hat + gamma_hat_1.iloc[int(n/2):] + (1 - pi_hat) * gamma_hat_0.iloc[int(n/2) : ]\n",
    "\n",
    "val_st, val_sderr = np.mean(gamma_hat_pi), np.std(gamma_hat_pi) / np.sqrt(len(gamma_hat_pi))\n",
    "val_st, val_sderr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "A technical note. Very small policy tree leaves make it hard to reliably evaluate policy values, in particular when the treatment is categorical with many levels. You can avoid small tree leaves increasing the `min.node.size` argument in `policy_tree`.\n",
    "\n",
    "\n",
    "[Possible edit here: talk about cross-validation?]\n",
    "\n",
    "\n",
    "## Case study\n",
    "\n",
    "Let's apply the methods above to our `welfare` dataset, as used in previous chapters.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read in data\n",
    "data = pd.read_csv(\"https://docs.google.com/uc?id=1kSxrVci_EUcSr_Lg1JKk1l7Xd5I9zfRC&export=download\")\n",
    "\n",
    "# Extract rows from data\n",
    "n_row, ncol = data.shape\n",
    "\n",
    "# ## NOTE: invert treatment and control, compared to the ATE and HTE chapters.\n",
    "\n",
    "data['w'] = 1 - data['w']\n",
    "\n",
    "# # Treatment is the wording of the question:\n",
    "# # 'does the the gov't spend too much on 'assistance to the poor' (control: 0)\n",
    "# # 'does the the gov't spend too much on \"welfare\"?' (treatment: 1)\n",
    "treatment = \"w\"\n",
    "\n",
    "# # Outcome: 1 for 'yes', 0 for 'no'\n",
    "outcome = \"y\"\n",
    "\n",
    "# # Additional covariates\n",
    "covariates = [\"age\", \"polviews\", \"income\", \"educ\", \"marital\", \"sex\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It's important to note that there are different types of \"heterogeneity\" in treatment effects. Sometimes the effect of a treatment is positive throughout, and what changes is the magnitude of the effect. In this case, we would still like to treat everyone. On the other hand, sometimes the treatment effect is positive for certain subgroups and negative for others. The latter is a more interesting scenario for policy learning. \n",
    "\n",
    "In this dataset, however, the effect seems to be mostly positive throughout. That is, i.e., most individuals respond \"yes\" more often when they are asked about \"welfare\" than about \"assistance to the poor\". To make the problem more interesting, we'll artificially modify the problem by introducing a cost of asking about welfare. This is just for illustration here, although there are natural examples in which treatment is indeed costly. Note in the code below how we subtract a cost of `.3` from the AIPW scores associated with treatment.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Prepare data\n",
    "x = data[covariates]\n",
    "y = data[outcome]\n",
    "w = data[treatment]\n",
    "cost = .3\n",
    "\n",
    "# Fit a policy tree on forest-based AIPW scores\n",
    "forest.tune(y, w, X=x, W=None)\n",
    "forest_aipw = forest.fit(y, w, X=x, W=None, cache_values=True)\n",
    "# time.sleep(5) \n",
    "\n",
    "gamma_hat_1, gamma_hat_0 = double_robust_score(forest_aipw, y, w, x)\n",
    "# time.sleep(5)\n",
    "\n",
    "# Substracting cost of treatment\n",
    "gamma_hat_1 -= cost\n",
    "gamma_mtrx = pd.DataFrame(\n",
    "    {\n",
    "        \"gamm_hat_0\" : gamma_hat_0,\n",
    "        \"gamm_hat_1\": gamma_hat_1\n",
    "    }\n",
    ")\n",
    "\n",
    "# Divide data into train and evaluation sets\n",
    "train = int(n_row * .8)\n",
    "\n",
    "# Fit policy on training subset\n",
    "policy = PolicyTree(\n",
    "    max_depth=2, honest=True, random_state=2\n",
    ").fit(x.iloc[ : train], gamma_mtrx.iloc[ : train])\n",
    "\n",
    "# Predicting leaves (useful later)\n",
    "pi_hat = policy.predict(x.iloc[train : ])\n",
    "leaf = policy.apply(x.iloc[train : ])\n",
    "num_leave = len(set(leaf))\n",
    "# policy.pre leaf by obsservacion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1800x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(25, 5))\n",
    "policy.plot(treatment_names = [\"Control\", \"Treatment\"])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Estimating the value of the learned policy. Note in the code below that we must subtract the cost of treatment.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Print value estimate [sample avg], and standard error\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.326752736539935, \\  0.00860428074467554\\right)$"
      ],
      "text/plain": [
       "(0.3267527365399351, 0.008604280744675535)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "a = pi_hat == 1\n",
    "\n",
    "y_test = y.iloc[train : ] \n",
    "w_test = w.iloc[train : ]\n",
    "\n",
    "print(\"Print value estimate [sample avg], and standard error\")\n",
    "\n",
    "# Print Value_estimate [sample avg] \n",
    "extr_val_sd(y_test, w_test, a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Estimate [AIPW], Value Estiamte and standar error\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.117586823970906, \\  0.00806894087325494\\right)$"
      ],
      "text/plain": [
       "(0.11758682397090611, 0.008068940873254937)"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Valid in both randomized and obs setting with unconf + overlap.\n",
    "### AIPW\n",
    "### Results from double_robust_score(): gamma_hat_1, gamma_hat_0\n",
    "gamma_hat_pi = pi_hat * (gamma_hat_1.iloc[train:]) + (1 - pi_hat) * gamma_hat_0.iloc[train : ]\n",
    "\n",
    "print(\"Estimate [AIPW], Value Estiamte and standar error\")\n",
    "## Print Estimate [AIPW}\n",
    "np.mean(gamma_hat_pi), np.std(gamma_hat_pi) / np.sqrt(len(gamma_hat_pi))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Testing whether the learned policy value is different from the value attained by the \"no-treatment\" policy.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Diference estimate [AIPW]:\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/latex": [
       "$\\displaystyle \\left( 0.0524440708684562, \\  0.00398411165396446\\right)$"
      ],
      "text/plain": [
       "(0.05244407086845618, 0.003984111653964457)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Only valid for randomized setting.\n",
    "c_1 = a & (w_test == 1)\n",
    "c_0 = np.logical_not(a) & (w_test == 0)\n",
    "\n",
    "y_main = pd.DataFrame(y_test)\n",
    "y_main['c_0'], y_main['c_1'] = c_0, c_1\n",
    "y_0 = y_main.loc[y_main['c_0'] == True]['y']\n",
    "y_1 = y_main.loc[y_main['c_1'] == True]['y']\n",
    "\n",
    "diff_estimate = (np.mean(y_1) - cost - np.mean(y_0)) * np.mean(a)\n",
    "diff_strerr = np.sqrt(np.var(y_1) / np.sum(c_1) * np.mean(a)**2 +  np.var(y_0) / np.sum(c_0) * np.mean(a)**2) \n",
    "diff_estimate, diff_strerr\n",
    "\n",
    "gamma_hat_pi_diff = gamma_hat_pi - gamma_hat_0\n",
    "diff_estimate = np.mean(gamma_hat_pi_diff)\n",
    "diff_strerr = np.std(gamma_hat_pi_diff) / np.sqrt(len(gamma_hat_pi_diff))\n",
    "\n",
    "print(\"Diference estimate [AIPW]:\")\n",
    "diff_estimate, diff_strerr"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Topics 1: Subgroups using learned policy\n",
    "\n",
    "The learned policy naturally induces interesting subgroups for which we expect the treatment effect to be different. With appropriate sample splitting, we can test that treatment effect is indeed different across \"regions\" defined by assignment under the learned policy,\n",
    "\n",
    "$$\n",
    "  H_0: \\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0)| \\hat{\\pi}(X_i) = 1] = \\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0)| \\hat{\\pi}(X_i) = 0].\n",
    "$$\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n"
     ]
    },
    {
     "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>index</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>w:C(pi_hat)[0]</td>\n",
       "      <td>-0.081289</td>\n",
       "      <td>0.017086</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>w:C(pi_hat)[1]</td>\n",
       "      <td>0.070428</td>\n",
       "      <td>0.013240</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            index     Coef.  Std.Err.\n",
       "2  w:C(pi_hat)[0] -0.081289  0.017086\n",
       "3  w:C(pi_hat)[1]  0.070428  0.013240"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## Olny from randomized settings\n",
    "\n",
    "## subset test data\n",
    "\n",
    "data_test = data.iloc[train : ]\n",
    "data_test['pi_hat'] = pi_hat\n",
    "\n",
    "## Formula\n",
    "fmla = outcome + \" ~ 0 + C(pi_hat) + w:C(pi_hat)\"\n",
    "\n",
    "ols = smf.ols(fmla, data=data_test).fit(cov_type='HC2')\n",
    "ols_coef = ols.summary2().tables[1].reset_index()\n",
    "ols_coef['Coef.'] = ols_coef['Coef.'] - cost\n",
    "ols_coef.iloc[2:4, 0:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "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>index</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "      <th>z</th>\n",
       "      <th>P&gt;|z|</th>\n",
       "      <th>[0.025</th>\n",
       "      <th>0.975]</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>C(pi_hat)[0]</td>\n",
       "      <td>-0.078724</td>\n",
       "      <td>0.016385</td>\n",
       "      <td>-4.804744</td>\n",
       "      <td>1.549493e-06</td>\n",
       "      <td>-0.110837</td>\n",
       "      <td>-0.046611</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>C(pi_hat)[1]</td>\n",
       "      <td>0.074598</td>\n",
       "      <td>0.012657</td>\n",
       "      <td>5.893816</td>\n",
       "      <td>3.773772e-09</td>\n",
       "      <td>0.049791</td>\n",
       "      <td>0.099406</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          index     Coef.  Std.Err.         z         P>|z|    [0.025  \\\n",
       "0  C(pi_hat)[0] -0.078724  0.016385 -4.804744  1.549493e-06 -0.110837   \n",
       "1  C(pi_hat)[1]  0.074598  0.012657  5.893816  3.773772e-09  0.049791   \n",
       "\n",
       "     0.975]  \n",
       "0 -0.046611  \n",
       "1  0.099406  "
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Valid in randomized settings and observational settings with unconfoundedness+overlap\n",
    "\n",
    "gamma_diff = gamma_hat_1 - gamma_hat_0\n",
    "\n",
    "### gamma_diff as y \n",
    "ga_df = pd.DataFrame({'gamma_diff': gamma_diff}).iloc[train : ]\n",
    "ga_df['pi_hat'] = pi_hat\n",
    "gam_fml = \"gamma_diff ~ 0 + C(pi_hat)\"\n",
    "ols = smf.ols(gam_fml, data = ga_df).fit(cov_type = \"HC2\").summary2().tables[1].reset_index()\n",
    "ols"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we learned a tree policy using the `policytree`, we can test whether treatment effects are different across leaves.\n",
    "\n",
    "$$\n",
    "  H_0: \\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0)| \\text{Leaf} = 1] = \\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0)| \\text{Leaf} = \\ell] \\qquad \\text{for }\\ell \\geq 2\n",
    "$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n"
     ]
    },
    {
     "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>index</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>w:C(leaf)[2]</td>\n",
       "      <td>0.252858</td>\n",
       "      <td>0.024942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>w:C(leaf)[3]</td>\n",
       "      <td>0.190775</td>\n",
       "      <td>0.057241</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>w:C(leaf)[5]</td>\n",
       "      <td>0.184534</td>\n",
       "      <td>0.025652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>w:C(leaf)[6]</td>\n",
       "      <td>0.370428</td>\n",
       "      <td>0.013240</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          index     Coef.  Std.Err.\n",
       "4  w:C(leaf)[2]  0.252858  0.024942\n",
       "5  w:C(leaf)[3]  0.190775  0.057241\n",
       "6  w:C(leaf)[5]  0.184534  0.025652\n",
       "7  w:C(leaf)[6]  0.370428  0.013240"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Only valid in randomized settings.\n",
    "fmla = outcome + \" ~ + C(leaf) + w:C(leaf)\"\n",
    "\n",
    "data_test['leaf'] = leaf\n",
    "ols = smf.ols(fmla, data=data_test).fit(cov_type = \"HC2\")\n",
    "ols_coef = ols.summary2().tables[1].reset_index()\n",
    "ols_coef.loc[ols_coef[\"index\"].str.contains(\":\")].iloc[:, 0:3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "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>index</th>\n",
       "      <th>Coef.</th>\n",
       "      <th>Std.Err.</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>C(leaf)[2]</td>\n",
       "      <td>-0.038142</td>\n",
       "      <td>0.023435</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>C(leaf)[3]</td>\n",
       "      <td>-0.110129</td>\n",
       "      <td>0.058267</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>C(leaf)[5]</td>\n",
       "      <td>-0.119484</td>\n",
       "      <td>0.024711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>C(leaf)[6]</td>\n",
       "      <td>0.074598</td>\n",
       "      <td>0.012657</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        index     Coef.  Std.Err.\n",
       "0  C(leaf)[2] -0.038142  0.023435\n",
       "1  C(leaf)[3] -0.110129  0.058267\n",
       "2  C(leaf)[5] -0.119484  0.024711\n",
       "3  C(leaf)[6]  0.074598  0.012657"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Valid in randomized settings and observational settings with unconfoundedness+overlap.\n",
    "ga_df['leaf'] = leaf\n",
    "fml_gm = \"gamma_diff ~ 0 + C(leaf)\"\n",
    "ols = smf.ols(fml_gm, data = ga_df).fit(cov_type = \"HC2\").summary2().tables[1].reset_index()\n",
    "ols.iloc[:, 0:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Finally, as we have done in previous chapters, we can check how covariate averages vary across subgroups. This time, the subgroups are defined by treatment assignment under the learned policy.\n",
    "\n",
    "$$\n",
    "  H_0: \\mathop{\\mathrm{E}}[X_{ij} | \\hat{\\pi}(X_i) = 1] = \\mathop{\\mathrm{E}}[X_{ij} | \\hat{\\pi}(X_i) = 0] \\qquad \\text{for each covariate }j\n",
    "$$\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "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>covariate</th>\n",
       "      <th>avg</th>\n",
       "      <th>stderr</th>\n",
       "      <th>scaling</th>\n",
       "      <th>variation</th>\n",
       "      <th>labels</th>\n",
       "      <th>pi_hat</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>C(pi_hat)[0]</th>\n",
       "      <td>age</td>\n",
       "      <td>46.6263</td>\n",
       "      <td>0.447695</td>\n",
       "      <td>0.158655</td>\n",
       "      <td>0.0205182</td>\n",
       "      <td>46.63\\n(0.45)</td>\n",
       "      <td>Control</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(pi_hat)[1]</th>\n",
       "      <td>age</td>\n",
       "      <td>47.3177</td>\n",
       "      <td>0.25541</td>\n",
       "      <td>0.841345</td>\n",
       "      <td>0.0205182</td>\n",
       "      <td>47.32\\n(0.26)</td>\n",
       "      <td>Treatment</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C(pi_hat)[0]</th>\n",
       "      <td>polviews</td>\n",
       "      <td>3.04583</td>\n",
       "      <td>0.0377259</td>\n",
       "      <td>0.158655</td>\n",
       "      <td>0.556402</td>\n",
       "      <td>3.05\\n(0.04)</td>\n",
       "      <td>Control</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             covariate      avg     stderr   scaling  variation  \\\n",
       "C(pi_hat)[0]       age  46.6263   0.447695  0.158655  0.0205182   \n",
       "C(pi_hat)[1]       age  47.3177    0.25541  0.841345  0.0205182   \n",
       "C(pi_hat)[0]  polviews  3.04583  0.0377259  0.158655   0.556402   \n",
       "\n",
       "                     labels     pi_hat  \n",
       "C(pi_hat)[0]  46.63\\n(0.45)    Control  \n",
       "C(pi_hat)[1]  47.32\\n(0.26)  Treatment  \n",
       "C(pi_hat)[0]   3.05\\n(0.04)    Control  "
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame()\n",
    "\n",
    "for var_name in covariates:\n",
    "    form2 = var_name + \" ~ 0 + C(pi_hat)\"\n",
    "    ols = smf.ols(formula=form2, data=data_test).fit(cov_type = 'HC2').summary2().tables[1].iloc[:, 0:2]\n",
    "    \n",
    "    nrow, ncol = ols.shape\n",
    "    \n",
    "    # Retrieve results\n",
    "    toget_index = ols[\"Coef.\"]\n",
    "    index = toget_index.index\n",
    "    cova1 = pd.Series(np.repeat(var_name,nrow), index = index, name = \"covariate\")\n",
    "    avg = pd.Series(ols[\"Coef.\"], name=\"avg\")\n",
    "    stderr = pd.Series(ols[\"Std.Err.\"], name = \"stderr\")\n",
    "    # ranking = pd.Series(np.arange(1,nrow+1), index = index, name = \"ranking\")\n",
    "    # ranking = pd.Series(np.array([\"Control\", \"Treatment\"]), index = index, name = \"pi_hat\"),\n",
    "    scaling = pd.Series(norm.cdf((avg - np.mean(avg))/np.std(avg)), index = index, name = \"scaling\")\n",
    "    # data2 = pd.DataFrame(data=X, columns= covariates)\n",
    "    variation1= np.std(avg) / np.std(data[var_name])\n",
    "    variation = pd.Series(np.repeat(variation1, nrow), index = index, name = \"variation\")\n",
    "    labels = pd.Series(round(avg,2).astype('str') + \"\\n\" + \"(\" + round(stderr, 2).astype('str') + \")\", index = index, name = \"labels\")\n",
    "    \n",
    "    # Tally up results\n",
    "    df1 = pd.DataFrame(data = [cova1, avg, stderr, scaling, variation, labels]).T\n",
    "    df = df.append(df1)\n",
    "\n",
    "\n",
    "df[\"pi_hat\"] = [\"Control\", \"Treatment\"]*len(covariates) \n",
    "df.head(3)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Average covariate values within each leaf')"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df1 = df.pivot(\"covariate\", \"pi_hat\", \"scaling\").astype(float)\n",
    "labels = df.pivot('covariate', 'pi_hat', 'labels').to_numpy()\n",
    "\n",
    "ax = plt.subplots(figsize=(10, 10))\n",
    "ax = sns.heatmap(\n",
    "    df1, annot=labels,\n",
    "    annot_kws={\"size\": 12, 'color':\"k\"},\n",
    "    fmt = '',\n",
    "    cmap = \"YlGn\",\n",
    "    linewidths=0\n",
    ")\n",
    "plt.tick_params( axis='y', labelsize=15, length=0, labelrotation=0)\n",
    "plt.tick_params( axis='x', labelsize=15, length=0, labelrotation=0)\n",
    "plt.xlabel(\"\", fontsize= 10)\n",
    "plt.ylabel(\"\")\n",
    "plt.title(\"Average covariate values within each leaf\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Topics 2: Learning with uncertain costs\n",
    "\n",
    "\n",
    "In the previous section, treatment costs were known and (just for simplicity of exposition) fixed across covariate values. However, there are situations in which costs are unknown and must be learned from the data as well. In such situations, we may interested only in policies that do not exceed a certain budget  in expectation.\n",
    "\n",
    "Here, we follow [Sun, Du, Wager (2021)](https://arxiv.org/abs/2103.11066) for how to deal with this issue. Their formulation is as follows. In potential outcome notation, each observation can be described by the tuple $(X_i, Y_i(0), Y_i(1), C_i(0), C_i(1))$, where the new pair $(C_i(0), C_i(1))$ represents costs that would be realized if individuals were assigned to control or treatment. Of course, in the data we only observe the tuple $(X_i, W_i, Y_i, C_i)$, where $C_i \\equiv C_i(W_i)$. We are interested in approximating the policy $\\pi_B^*$ that maximizes the gain from treating relative to not treating anyone while kee\\\\ping the average relative cost bounded by some known budget $B$,\n",
    "\n",
    "$$\n",
    "  \\\\\\pi_B^*(x) := \\arg\\max \\mathop{\\mathrm{E}}[Y(\\pi(X_i))] - \\mathop{\\mathrm{E}}[Y_i(0)]  \\quad \\text{such that} \\quad \\mathop{\\mathrm{E}}[C_i(\\pi(X_i)) - C_i(0)] \\leq B.\n",
    "$$\n",
    "\n",
    "This paper demonstrates that the optimal policy has the following structure. First, we can order observations in decreasing order according to the following priority ranking,\n",
    "\n",
    "$$ \n",
    "  \\rho(x) := \n",
    "    \\frac{\\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0) | X_i = x]}\n",
    "         {\\mathop{\\mathrm{E}}[C_i(1) - C_i(0) | X_i = x]}.\n",
    "$$ (rho)\n",
    "\n",
    "Then, we assign treatment in decreasing order {eq}`rho` until we either treat everyone with positive $\\rho(x)$ or the budget is met. The intuition is that individuals for which {eq}`rho` is high have a high expected treatment effect relative to cost, so by assigning them first we obtain a cost-effective policy. We stop once there's no one else for which treatment is expected to be positive or we run out of resources.\n",
    "\n",
    "To obtain estimates $\\hat{\\rho}$ of {eq}`rho` from the data, we have two options. The first is to estimate the numerator $\\widehat{\\tau}(x) = \\mathop{\\mathrm{E}}[Y_i(1) - Y_i(0) |X_i = x]$ and the denominator $\\hat{\\widehat{\\Gamma}}(x) = \\mathop{\\mathrm{E}}[C_i(1) - C_i(0) |X_i = x]$ separately, in a manner analogous to what we saw in the HTE chapter, and compute their ratio, producing the estimate $\\hat{\\rho}(x) = \\widehat{\\tau}(x) / \\hat{\\widehat{\\Gamma}}(x)$. We'll see a second option below. \n",
    "\n",
    "Let's put the above into practice. For illustration, we will generate random costs for our data. We'll assume that the costs of treatment are drawn from a conditionally Exponential distribution, and that there are no costs for no treating.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating random costs.\n",
    "nrow, ncol = data.shape\n",
    "cond = [data['w'] == 1]\n",
    "# do_it = [12]\n",
    "do_it = [np.random.uniform(0, 1, 1)]\n",
    "data['cost'] = C = np.select(cond, do_it, 0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The next snippet compares two kinds of policies. An \"ignore costs\" policy which, as the name suggests, orders individuals by $\\hat{\\tau}$ only without taking costs into account; and the \"ratio\" policy in which the numerator and denominator of {eq}`rho` are estimated separately. The comparison is made via a **cost curve** that compares the cumulative benefit of treatment with its cumulative cost (both normalized to 1), for all possible budgets at once. More cost-effective policies hug the left corner of the graph more tightly, kee\\\\ping away from the 45-degree line.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Assuming that the assignment probability is known.\n",
    "# If these are not known, they must be estimated from the data as usual.\n",
    "e = 0.5  \n",
    "\n",
    "y = data[outcome]\n",
    "w = data[treatment]\n",
    "x = data[covariates]\n",
    "\n",
    "# Sample splitting. \n",
    "# Note that we can't simply rely on out-of-bag observations here.\n",
    "train = int(nrow / 2)\n",
    "\n",
    "# IPW-based estimates of (normalized) treatment and cost\n",
    "n_test = int(nrow - nrow / 2)\n",
    "treatment_ipw = 1 / n_test * (w.iloc[train : ] / e - (1 - w.iloc[train : ]) / (1 - e)) * y.iloc[train : ]\n",
    "cost_ipw = 1 / n_test * w.iloc[train : ] / e * data['cost'].iloc[train : ]\n",
    "forest = causal_forest(\n",
    "    model_t=RegressionForest(),\n",
    "    model_y=RegressionForest(),\n",
    "    n_estimators=200, min_samples_leaf=5,\n",
    "    max_depth=50, verbose=0, random_state=1\n",
    ")\n",
    "forest.tune(y.iloc[:train], w.iloc[:train], X=x.iloc[:train], W=None)\n",
    "\n",
    "# # Compute predictions on test set\n",
    "tau_forest = forest.fit(y.iloc[: train], w.iloc[: train], X=x.iloc[: train], W=None)\n",
    "tau_hat = tau_forest.effect(x.iloc[train : ])\n",
    "\n",
    "# Estimating the denominator.\n",
    "# Because costs for untreated observations are known to be zero, we're only after E[C(1)|X].\n",
    "# Under unconfoundedness, this can be estimated by regressing C on X using only the treated units.\n",
    "gamm_forest = forest.fit(data['cost'].iloc[: train], w.iloc[: train], X=x.iloc[: train], W=None)\n",
    "gamm_hat = gamm_forest.effect(x.iloc[train : ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Rankings\n",
    "rank_ignore_cost = (-tau_hat).argsort()\n",
    "rank_ratio = (-gamm_hat).argsort()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create w_hat_test data_frame\n",
    "ipw = pd.DataFrame({'treatment_ipw': treatment_ipw, \"cost_ipw\": cost_ipw})\n",
    "ipw['rank_ignore_cost'] = rank_ignore_cost\n",
    "ipw['rank_ratio'] = rank_ratio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Cumulative benefit and cost of treatment (normalized) for a policy that ignores costs.\n",
    "treatment_value_ignore_cost = np.cumsum(ipw.sort_values(\"rank_ignore_cost\")['treatment_ipw']) / np.sum(ipw['treatment_ipw'])\n",
    "treatment_cost_ignore_cost = np.cumsum(ipw.sort_values(\"rank_ignore_cost\")['cost_ipw']) / np.sum(ipw['cost_ipw'])\n",
    "\n",
    "# Cumulative benefit and cost of treatment (normalized) for a policy that uses the ratio, estimated separately.\n",
    "treatment_value_ratio = np.cumsum(ipw.sort_values(\"rank_ratio\")['treatment_ipw']) / np.sum(ipw['treatment_ipw'])\n",
    "treatment_cost_ratio = np.cumsum(ipw.sort_values(\"rank_ratio\")['cost_ipw']) / np.sum(ipw['cost_ipw'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "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": [
    "plt.plot(treatment_cost_ignore_cost, treatment_value_ignore_cost, '#0d5413', label='Ignoring costs')\n",
    "plt.plot(treatment_cost_ratio, treatment_value_ratio, '#7c730d', label='Ratio')\n",
    "plt.title(\"Cost Curves\")\n",
    "plt.xlabel(\"(Normalized) cumulative cost\")\n",
    "plt.ylabel(\"(Normalized) cumulative value\")\n",
    "plt.plot([0, 1], [0, 1], color = 'black', linewidth = 1.5, linestyle = \"dotted\")\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To read this graph, we consider a point on the horizontal axis, representing a possible (normalized) budget constraint. At that point, whichever policy is higher is more cost-effective.\n",
    "\n",
    "As the authors note, we can also estimate {eq}`rho` in a second way. First, they note that, under overlap and the following extended unconfoudedness assumption\n",
    "\n",
    "$$\n",
    "  \\{Y_i(0), Y_i(1), C_i(1), C_i(0) \\perp W_i | X_i \\},\n",
    "$$\n",
    "\n",
    "we can rewrite {eq}`rho` as\n",
    "\n",
    "$$\n",
    "  \\rho(x) := \n",
    "    \\frac{\\text{Cov}[Y_i, W_i | X_i = x]}\n",
    "         {\\text{Cov}[C_i, W_i | X_i = x]}.\n",
    "$$ (rho-iv)\n",
    "\n",
    "As readers with a little more background in causal inference may note, {eq}`rho-iv` coincides with the definition of the conditional local average treatment effect (LATE) if we _were_ to take $W_i$ as an \"instrumental variable\" and $C_i$ as the \"treatment\". In fact, instrumental variable methods require different assumptions, so the connection with instrumental variables is tenuous (see the paper for details), but mechanically {eq}`rho-iv` provides us with an estimation procedure: we can use any method used to estimate conditional LATE to produce an estimate $\\hat{\\rho}$.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Estimating rho(x) directly via instrumental forests.\n",
    "# In observational settings, remove the argument W.hat.\n",
    "i_f = instrumental_forest().fit(x.iloc[:train], w.iloc[:train], y.iloc[:train], Z = data['cost'].iloc[:train])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Predict and compute and estimate of the ranking on a test set.\n",
    "rho_iv = i_f.predict(x.iloc[train : ])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create objects\n",
    "ipw['rho_iv']  = rho_iv\n",
    "ipw['rank_iv'] = (-rho_iv).argsort()\n",
    "\n",
    "treatment_valu_iv = np.cumsum(ipw.sort_values(\"rank_iv\")['treatment_ipw']) / sum(ipw['treatment_ipw'])\n",
    "treatment_cost_iv = np.cumsum(ipw.sort_values(\"rank_iv\")[\"cost_ipw\"]) / sum(ipw['cost_ipw'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "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": [
    "plt.plot(treatment_cost_ignore_cost, treatment_value_ignore_cost, '#0d5413', label='Ignoring costs')\n",
    "plt.plot(treatment_cost_ratio, treatment_value_ratio, '#7c730d', label='Ratio')\n",
    "plt.plot(treatment_cost_iv, treatment_valu_iv, \"#af1313\", label = \"Sun. Du. Wager (2021)\")\n",
    "plt.title(\"Cost Curves\")\n",
    "plt.xlabel(\"(Normalized) cumulative cost\")\n",
    "plt.ylabel(\"(Normalized) cumulative value\")\n",
    "plt.legend()\n",
    "plt.plot([0, 1], [0, 1], color = 'black', linewidth = 2)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this example, both the “direct ratio” and the solution based on instrumental forests have similar performance. This isn’t always the case. When the ratio $\\rho(x)$ is simpler relative to $\\tau(x)$ and $\\gamma(x)$, the solution based on instrumental forests may perform better since it is estimating $\\rho(x)$ directly, where the “direct ratio” solution needs to estimate the more complicated objects $\\tau(x)$ and $\\gamma(x)$ separately. At a high level, we should expect $\\rho(x)$ to be relatively simpler when there is a strong relationship between $\\tau(x)$ and $\\gamma(x)$. Here, our simulated costs seem to be somewhat related to CATE (see the plot below), but perhaps not strongly enough to make the instrumental forest solution noticeably better than the one based on ratios."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Esimated CATE (normalized)')"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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\n",
    "sns.scatterplot(x = gamm_hat, y = tau_hat)\n",
    "plt.xlabel(\"Estimated cost (normalized)\")\n",
    "plt.ylabel(\"Esimated CATE (normalized)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The different policies can be compared by the area between the curves they trace and the 45-degree line, with higher values indicating better policies.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [],
   "source": [
    "t_c_i_c = pd.Series(np.array(treatment_cost_ignore_cost))\n",
    "t_c_r = pd.Series(np.array(treatment_cost_ratio))\n",
    "t_c_iv = pd.Series(np.array(treatment_cost_iv))\n",
    "ignore = np.sum((np.array(treatment_value_ignore_cost) - t_c_i_c) * (t_c_i_c - t_c_i_c.shift(1)))\n",
    "ratio = np.sum((np.array(treatment_value_ratio) - t_c_r) * (t_c_r - t_c_r.shift(1)))\n",
    "iv = np.sum((np.array(treatment_valu_iv) - t_c_iv) * (t_c_iv - t_c_iv.shift(1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "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>ig</th>\n",
       "      <th>ratio</th>\n",
       "      <th>iv</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-0.004254</td>\n",
       "      <td>-0.001414</td>\n",
       "      <td>-0.007943</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ig     ratio        iv\n",
       "0 -0.004254 -0.001414 -0.007943"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame({\n",
    "    \"ig\": ignore,\n",
    "     \"ratio\": ratio,\n",
    "    \"iv\" : iv\n",
    "}, index = [0])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Further reading\n",
    "\n",
    "The presentation of parametric policies was largely based on Athey and Wager (Econometrica, 2021). A slightly more accessible version of some of the material in the published version can be found in an earlier [ArXiv version](https://arxiv.org/abs/1702.02896v1) of the same paper. Policy comparisons via cost curves can also be found in [Imai and Li (2019)](https://arxiv.org/pdf/1905.05389.pdf).\n"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": "",
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  },
  "vscode": {
   "interpreter": {
    "hash": "f20a34ee552d324c2ef46084ba449ea17833fb0de8b15cca223d94b97763f285"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}