27 - Control Scheme vs Estimator Variance.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "\n",
        "import statsmodels.api as sm\n",
        "\n",
        "from tqdm import tqdm\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "plt.style.use('fivethirtyeight')"
      ],
      "metadata": {
        "id": "WTrLsEbqJqnf"
      },
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "id": "1eXEgq1U11Gc"
      },
      "outputs": [],
      "source": [
        "def run_experiment(controls, n=50000):\n",
        "  # np.random.seed(45)\n",
        "  N_SAMPLES = n\n",
        "\n",
        "  # Generate X\n",
        "  age = np.random.randint(18, 65, N_SAMPLES)\n",
        "  re74 = np.random.randint(20000, 60000, N_SAMPLES)\n",
        "  hisp = np.random.choice(a=[0,1], size=N_SAMPLES)\n",
        "  black = np.random.choice(a=[0,1], size=N_SAMPLES)\n",
        "\n",
        "  re75 = re74 + np.random.randn(N_SAMPLES)\n",
        "  married = ((1 / (1 + np.exp(- (3 - (age*0.05) - (re75*0.00002) + np.random.randn(N_SAMPLES)))))>0.5).astype(int)\n",
        "  educ = -hisp + np.random.randint(7, 14, N_SAMPLES)\n",
        "  nodegr = ((1 / (1 + np.exp(- (1 - (age*0.05) - (educ*0.02) + np.random.randn(N_SAMPLES)))))>0.5).astype(int)\n",
        "  treat = ((1/(1+np.exp(nodegr + np.random.randn(N_SAMPLES))))>0.5).astype(int)\n",
        "\n",
        "  # Compute Y\n",
        "  re78 = age - 5*nodegr + 8*treat + educ - 4*black + np.random.randn(N_SAMPLES)\n",
        "\n",
        "  data = pd.DataFrame({\n",
        "  \"treat\": treat,\n",
        "  \"age\" : age,\n",
        "  \"re74\" :re74,\n",
        "  \"hisp\" : hisp,\n",
        "  \"black\": black,\n",
        "  \"re75\": re75,\n",
        "  \"married\": married,\n",
        "  \"educ\": educ,\n",
        "  \"nodegr\": nodegr\n",
        "  })\n",
        "\n",
        "  # Statsmodel models require us to add constant\n",
        "  vars = controls#[\"treat\", \"nodegr\"]#, \"age\",\"educ\", \"black\"]\n",
        "  X = data[vars]\n",
        "  X = sm.add_constant(X)\n",
        "\n",
        "  # Instantiate the model and fit it\n",
        "  model = sm.OLS(re78, X)\n",
        "  fitted_model = model.fit()\n",
        "\n",
        "  # Print results summary\n",
        "  # print(fitted_model.summary())\n",
        "  return fitted_model.params['treat']"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "var_sets = {\n",
        "    'full': [\"treat\", \"nodegr\", \"age\",\"educ\", \"black\"],\n",
        "     'minimal': [\"treat\", \"nodegr\"]\n",
        "}\n",
        "\n",
        "results = []\n",
        "\n",
        "for i in tqdm(range(500)):\n",
        "  for k, v in var_sets.items():\n",
        "    result = run_experiment(controls=v, n=10000)\n",
        "    results.append((k, result))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LpfRxCUI16Ix",
        "outputId": "f515be82-5ac1-4c76-9ba7-d97fe698e4ba"
      },
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "100%|██████████| 500/500 [00:17<00:00, 28.59it/s]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df = pd.DataFrame(results)"
      ],
      "metadata": {
        "id": "5ghcaPPtHqbJ"
      },
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.hist(df[df[0]=='full'][1], alpha=.4, label='full', bins=10)\n",
        "plt.hist(df[df[0]=='minimal'][1], alpha=.4, label='minimal', bins=20)\n",
        "plt.axvline(8, ls='--', color='black', lw=1, label='true')\n",
        "plt.legend()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 467
        },
        "id": "gxXgnvH7JP-7",
        "outputId": "8b6633bb-b618-4b1e-e37e-d26923f3e6cb"
      },
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7c8c909918a0>"
            ]
          },
          "metadata": {},
          "execution_count": 57
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "gJ7lxop7JnC9"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}