{ "cells": [ { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "# Python: Choice of learners\n", "\n", "This notebook contains some practical recommendations to choose the right learner and evaluate different learners for the corresponding nuisance components.\n", "This notebook mainly highlights the differences in using different learners, i.e. linear or tree-based methods. Generally, we recommend to tune hyperparameters for the chosen learners, see [Example Gallery](https://docs.doubleml.org/stable/examples/index.html).\n", "\n", "For the example, we will work with a IRM, but all of the important components are directly usable for all other models, too.\n", "\n", "To be able to compare the properties of different learners, we will start by setting the true treatment parameter to zero, fix some other parameters of the data generating process and generate several datasets \n", "to obtain some information about the distribution of the estimators." ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "import numpy as np\n", "import doubleml as dml\n", "\n", "from doubleml.irm.datasets import make_irm_data\n", "\n", "import warnings\n", "\n", "warnings.filterwarnings(\"ignore\", message=\"The estimated nu2 for d is not positive\")\n", "\n", "theta = 0\n", "n_obs = 500\n", "dim_x = 5\n", "n_rep = 200\n", "\n", "np.random.seed(42)\n", "datasets = []\n", "for i in range(n_rep):\n", " data = make_irm_data(theta=theta, n_obs=n_obs, dim_x=dim_x, \n", " R2_d=0.8, R2_y=0.8, return_type='DataFrame')\n", " datasets.append(dml.DoubleMLData(data, 'y', 'd'))" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "## Comparing different learners\n", "For simplicity, we will restrict ourselves to the comparison of two different types and evaluate a learner of linear type and a tree based estimator for each nuisance component (with default hyperparameters)." ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "from sklearn.linear_model import LinearRegression, LogisticRegressionCV\n", "from sklearn.ensemble import GradientBoostingRegressor, GradientBoostingClassifier\n", "from sklearn.base import clone\n", "\n", "reg_learner_1 = LinearRegression()\n", "reg_learner_2 = GradientBoostingRegressor()\n", "class_learner_1 = LogisticRegressionCV()\n", "class_learner_2 = GradientBoostingClassifier()\n", "\n", "learner_list = [{'ml_g': reg_learner_1, 'ml_m': class_learner_1},\n", " {'ml_g': reg_learner_2, 'ml_m': class_learner_1},\n", " {'ml_g': reg_learner_1, 'ml_m': class_learner_2},\n", " {'ml_g': reg_learner_2, 'ml_m': class_learner_2}]" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "In all combinations, we now can try to evaluate four different IRM models. To make the comparison fair, we will apply all different models to the same cross-fitting samples (usually this should not matter, we only consider this here to get slightly cleaner comparison)." ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "### Standard approach\n", "\n", "At first, we will look at the most straightforward approach using the inbuild nuisance losses. The `nuisance_loss` attribute contains the out-of-sample RMSE or Log Loss for the nuisance functions. We will save all RMSEs and the corresponding treatment estimates for all combinations of learners over all repetitions." ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Processing: 100.0 %\n", "Coverage: [0.935 0.65 0.975 0.94 ]\n" ] } ], "source": [ "from doubleml.utils import DoubleMLResampling\n", "\n", "coefs = np.full(shape=(n_rep, len(learner_list)), fill_value=np.nan)\n", "loss_ml_m = np.full(shape=(n_rep, len(learner_list)), fill_value=np.nan)\n", "loss_ml_g0 = np.full(shape=(n_rep, len(learner_list)), fill_value=np.nan)\n", "loss_ml_g1 = np.full(shape=(n_rep, len(learner_list)), fill_value=np.nan)\n", "\n", "coverage = np.full(shape=(n_rep, len(learner_list)), fill_value=np.nan)\n", "\n", "for i_rep in range(n_rep):\n", " print(f\"\\rProcessing: {round((i_rep+1)/n_rep*100, 3)} %\", end=\"\")\n", " dml_data = datasets[i_rep]\n", " # define the sample splitting\n", " smpls = DoubleMLResampling(n_folds=5, n_rep=1, n_obs=n_obs, stratify=dml_data.d).split_samples()\n", " \n", " for i_learners, learners in enumerate(learner_list):\n", " np.random.seed(42)\n", " dml_irm = dml.DoubleMLIRM(dml_data,\n", " ml_g=clone(learners['ml_g']),\n", " ml_m=clone(learners['ml_m']),\n", " draw_sample_splitting=False)\n", " dml_irm.set_sample_splitting(smpls)\n", " dml_irm.fit(n_jobs_cv=5)\n", "\n", " coefs[i_rep, i_learners] = dml_irm.coef[0]\n", " loss_ml_m[i_rep, i_learners] = dml_irm.nuisance_loss['ml_m'][0][0]\n", " loss_ml_g0[i_rep, i_learners] = dml_irm.nuisance_loss['ml_g0'][0][0]\n", " loss_ml_g1[i_rep, i_learners] = dml_irm.nuisance_loss['ml_g1'][0][0]\n", "\n", " confint = dml_irm.confint()\n", " coverage[i_rep, i_learners] = (confint['2.5 %'].iloc[0] <= theta) & (confint['97.5 %'].iloc[0] >= theta)\n", "\n", "print(f'\\nCoverage: {coverage.mean(0)}')" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "Next, let us take a look at the corresponding results" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "import pandas as pd\n", "\n", "colnames = ['Linear + Logit','Boost + Logit', 'Linear + Boost', 'Boost + Boost']\n", "\n", "df_coefs = pd.DataFrame(coefs, columns=colnames)\n", "df_ml_m = pd.DataFrame(loss_ml_m, columns=colnames)\n", "df_ml_g0 = pd.DataFrame(loss_ml_g0, columns=colnames)\n", "df_ml_g1 = pd.DataFrame(loss_ml_g1, columns=colnames)" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import seaborn as sns\n", "import matplotlib.pyplot as plt\n", "\n", "fig, axes = plt.subplots(2, 2, figsize=(15, 5))\n", "fig.suptitle('Learner Comparison')\n", "\n", "\n", "sns.kdeplot(data=df_coefs,ax=axes[0][0], fill=True, alpha=0.3)\n", "sns.kdeplot(data=df_ml_m, ax=axes[0][1], fill=True, alpha=0.3, legend=False)\n", "sns.kdeplot(data=df_ml_g0, ax=axes[1][0], fill=True, alpha=0.3, legend=False)\n", "sns.kdeplot(data=df_ml_g1, ax=axes[1][1], fill=True, alpha=0.3)\n", "\n", "axes[0][0].title.set_text('Estimated Parameter')\n", "axes[0][1].title.set_text('Log Loss ml_m')\n", "axes[1][0].title.set_text('RMSE ml_g0')\n", "axes[1][1].title.set_text('RMSE ml_g1')\n", "\n", "plt.subplots_adjust(left=0.1,\n", " bottom=0.1, \n", " right=0.9, \n", " top=0.9, \n", " wspace=0.4, \n", " hspace=0.4)\n", "plt.show()" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "We can now easily observe that in this setting, the linear learners are able to approximate the corresponding nuisance functions better than the boosting algorithm (as should be expected since the data is generated accordingly).\n", "\n", "Let us take a look at what would have happend if a each repetition for each nuisance element, we would have selected the learner with smallest out-of-sample loss (in our example this corresponds to minimizing the product of losses). \n", "Remark that we cannot select different learners for `ml_g0` and `ml_g1`." ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "(array([0, 2]), array([194, 6]))" ] }, "execution_count": 6, "metadata": {}, "output_type": "execute_result" } ], "source": [ "selected_learners = (loss_ml_m * (loss_ml_g0 + loss_ml_g1)).argmin(axis=1)\n", "np.unique(selected_learners, return_counts=True)" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "Most of the time, we will use linear learners for both nuisance elements. Sometimes the tree-based estimator is chosen for the propensity score `ml_m`. \n", "Let us compare which learners, how the estimated coefficients would have performed with the selected learners." ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Coverage of selected learners: 0.94\n" ] }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "print(f'Coverage of selected learners: {np.mean(np.array([coverage[i_rep, selected_learners[i_rep]] for i_rep in range(n_rep)]))}')\n", "\n", "selected_coefs = np.array([coefs[i_rep, selected_learners[i_rep]] for i_rep in range(n_rep)])\n", "df_coefs['Selected'] = selected_coefs\n", "sns.kdeplot(data=df_coefs, fill=True, alpha=0.3)\n", "plt.show()\n" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "This procedure will be generally valid as long as we do not compare a excessively large number of different learners." ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "### Custom evaluation metrics\n", "\n", "If one wants to evaluate a learner based on some other metric/loss it is possible to use the inbuilt `evaluate_learners()` method.\n", "Without further arguments this will default to the RMSE for all nuisance components and result in the same output as the `nuisance_loss` attribute for regressors." ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "{'ml_g0': array([[1.02699695]]), 'ml_g1': array([[1.06270135]]), 'ml_m': array([[0.34980811]])}\n", "{'ml_g0': array([[1.02699695]]), 'ml_g1': array([[1.06270135]]), 'ml_m': array([[0.39186467]])}\n" ] } ], "source": [ "print(dml_irm.evaluate_learners())\n", "print(dml_irm.nuisance_loss)" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "To evaluate a self-defined metric, the user has to hand over a callable. In this example, we define the mean absolute deviation as an error metric.\n", "\n", "Remark that the metric should be able to handle `nan` values, since e.g. in the IRM model the learner `ml_g` is used to onto two different subsamples. As a result, we have two different nuisance components for\n", "\n", "$$\n", "\\begin{aligned}\n", "g_0(x) &= \\mathbb{E}[Y|X=x, D=0] \\\\\n", "g_1(x) &= \\mathbb{E}[Y|X=x, D=1]\n", "\\end{aligned}\n", "$$\n", "\n", "which are both fitted with the learner `ml_g`. Of course, we can only observe the target value for $g_0(x)$ if $D=0$ and vice versa, resulting in `nan` values for all other observations." ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "{'ml_g0': array([[0.82329138]]),\n", " 'ml_g1': array([[0.85402594]]),\n", " 'ml_m': array([[0.20167273]])}" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from sklearn.metrics import mean_absolute_error\n", "\n", "def mae(y_true, y_pred):\n", " subset = np.logical_not(np.isnan(y_true))\n", " return mean_absolute_error(y_true[subset], y_pred[subset])\n", "\n", "dml_irm.evaluate_learners(metric=mae)" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "Another option is to access the out-of-sample predictions and target values for the nuisance elements via the `nuisance_targets` and `predictions` attributes." ] }, { "cell_type": "code", "execution_count": 10, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "(500, 1, 1)\n", "(500, 1, 1)\n" ] } ], "source": [ "print(dml_irm.nuisance_targets['ml_g1'].shape)\n", "print(dml_irm.predictions['ml_g1'].shape)" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "For most models minimizing the RMSE for each learner should result in improved performance as the theoretical backbone of the DML Framework is build on $\\ell_2$-convergence rates for the nuisance estimates ([Chernozhukov et al. (2018)](https://doi.org/10.1111/ectj.12097)). But for some models (e.g. classification learners) it might be helpful to further check other error metrics (e.g. as in [scikit-learn](https://scikit-learn.org/stable/modules/model_evaluation.html#)) to gain a overview whether the nuisance function can be approximated sufficiently well. Specifically, for binary classifications the log loss is a common and stable choice as it is also a calibrated metric. \n", "\n", "Of course, if one has some prior knowledge on functional form assumptions (e.g. linearity as in the IRM example above) using these learners will usually improve the performance of the estimator and might speed up computation time." ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "## Computation time\n", "\n", "The choice of the learner has a huge impact on the computation time of the DoubleML models. As the largest part of the computation time is usually used to train the learners for the nuisance components, some clever choices of learners and hyperparameters can speed up the computation time. \n", "\n", "Resourcewise, most implementations support the `n_jobs_cv` argument, which can parallelize the k-fold estimation and might speed up the calculation nearly up to $k$-times if the resources are available." ] }, { "cell_type": "code", "execution_count": 11, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Time without parallelization of crossfitting: 4.3593 seconds\n", "Time with parallelization of crossfitting: 0.8996 seconds\n", "Speedup of factor 4.85\n" ] } ], "source": [ "from sklearn.ensemble import RandomForestRegressor, RandomForestClassifier\n", "from time import perf_counter\n", "\n", "np.random.seed(42)\n", "n_obs = 1000\n", "dml_data = dml.DoubleMLData(make_irm_data(theta=0, n_obs=n_obs, dim_x=20, return_type='DataFrame'), 'y', 'd')\n", "\n", "# define the sample splitting\n", "smpls = DoubleMLResampling(n_folds=5, n_rep=1, n_obs=n_obs, stratify=dml_data.d).split_samples()\n", "\n", "dml_irm = dml.DoubleMLIRM(dml_data,\n", " ml_g=RandomForestRegressor(),\n", " ml_m=RandomForestClassifier(),\n", " draw_sample_splitting=False)\n", "dml_irm.set_sample_splitting(smpls)\n", "\n", "np.random.seed(42)\n", "t_1_start = perf_counter()\n", "dml_irm.fit()\n", "t_1_stop = perf_counter()\n", "print(f'Time without parallelization of crossfitting: {round(t_1_stop - t_1_start, 4)} seconds')\n", "\n", "np.random.seed(42)\n", "t_2_start = perf_counter()\n", "dml_irm.fit(n_jobs_cv=5)\n", "t_2_stop = perf_counter()\n", "print(f'Time with parallelization of crossfitting: {round(t_2_stop - t_2_start, 4)} seconds')\n", "print(f'Speedup of factor {round((t_1_stop - t_1_start) / (t_2_stop - t_2_start), 2)}')" ] }, { "attachments": {}, "cell_type": "markdown", "metadata": {}, "source": [ "Other more helpful ways to improve computation time will largly depend on the implemented learner. Of course linear learners are quite fast, but if no functional form restrictions are known Boosting or Random Forest might be better default options to saveguard against wrong model assumptions. Especially Boosting performs very well as a default option for tabular data. As a general recommendation all popular Boosting frameworks (XGBoost, Lightgbm, Catboost, etc.) should improve computation time.\n", "But this might vary heavily with the number of features in your dataset.\n", "Let us compare the computation time with Boosting and Random Forest (we increase the sample size and the number of features)." ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Time without RandomForest (Scikit-Learn): 8.8445 seconds\n", "Time with XGBoost: 1.8681 seconds\n", "Speedup of factor 4.73\n", "Time with LightGBM: 0.4814 seconds\n", "Speedup of factor 18.37\n" ] } ], "source": [ "from xgboost import XGBClassifier, XGBRegressor\n", "from lightgbm import LGBMClassifier, LGBMRegressor\n", "\n", "np.random.seed(42)\n", "n_obs = 1000\n", "dml_data = dml.DoubleMLData(make_irm_data(theta=0, n_obs=n_obs, dim_x=50, return_type='DataFrame'), 'y', 'd')\n", "\n", "# define the sample splitting\n", "smpls = DoubleMLResampling(n_folds=5, n_rep=1, n_obs=n_obs, stratify=dml_data.d).split_samples()\n", "\n", "np.random.seed(42)\n", "t_1_start = perf_counter()\n", "dml_irm = dml.DoubleMLIRM(dml_data,\n", " ml_g=RandomForestRegressor(),\n", " ml_m=RandomForestClassifier(),\n", " draw_sample_splitting=False)\n", "dml_irm.set_sample_splitting(smpls)\n", "dml_irm.fit()\n", "t_1_stop = perf_counter()\n", "print(f'Time without RandomForest (Scikit-Learn): {round(t_1_stop - t_1_start, 4)} seconds')\n", "\n", "np.random.seed(42)\n", "t_2_start = perf_counter()\n", "dml_irm = dml.DoubleMLIRM(dml_data,\n", " ml_g=XGBRegressor(),\n", " ml_m=XGBClassifier(),\n", " draw_sample_splitting=False)\n", "dml_irm.set_sample_splitting(smpls)\n", "dml_irm.fit()\n", "t_2_stop = perf_counter()\n", "print(f'Time with XGBoost: {round(t_2_stop - t_2_start, 4)} seconds')\n", "print(f'Speedup of factor {round((t_1_stop - t_1_start) / (t_2_stop - t_2_start), 2)}')\n", "\n", "np.random.seed(42)\n", "t_3_start = perf_counter()\n", "dml_irm = dml.DoubleMLIRM(dml_data,\n", " ml_g=LGBMRegressor(verbose=-1),\n", " ml_m=LGBMClassifier(verbose=-1),\n", " draw_sample_splitting=False)\n", "dml_irm.set_sample_splitting(smpls)\n", "dml_irm.fit()\n", "t_3_stop = perf_counter()\n", "print(f'Time with LightGBM: {round(t_3_stop - t_3_start, 4)} seconds')\n", "print(f'Speedup of factor {round((t_1_stop - t_1_start) / (t_3_stop - t_3_start), 2)}')" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "## References\n", "\n", "Bach, P., Schacht, O., Chernozhukov, V., Klaassen, S., & Spindler, M. (2024, March). Hyperparameter Tuning for Causal Inference with Double Machine Learning: A Simulation Study. In Causal Learning and Reasoning (pp. 1065-1117). PMLR." ] } ], "metadata": { "kernelspec": { "display_name": ".venv", "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": "3.12.3" }, "orig_nbformat": 4 }, "nbformat": 4, "nbformat_minor": 2 }