Estimation and Inference of Heterogeneous Treatment Effects using Random Forests (2015-10-14T00:00:00.000000Z)

TL;DR

This is the first set of results that allows any type of random forest, including classification and regression forests, to be used for provably valid statistical inference and is found to be substantially more powerful than classical methods based on nearest-neighbor matching.

Abstract

ABSTRACT Many scientific and engineering challenges—ranging from personalized medicine to customized marketing recommendations—require an understanding of treatment effect heterogeneity. In this article, we develop a nonparametric causal forest for estimating heterogeneous treatment effects that extends Breiman’s widely used random forest algorithm. In the potential outcomes framework with unconfoundedness, we show that causal forests are pointwise consistent for the true treatment effect and have an asymptotically Gaussian and centered sampling distribution. We also discuss a practical method for constructing asymptotic confidence intervals for the true treatment effect that are centered at the causal forest estimates. Our theoretical results rely on a generic Gaussian theory for a large family of random forest algorithms. To our knowledge, this is the first set of results that allows any type of random forest, including classification and regression forests, to be used for provably valid statistical inference. In experiments, we find causal forests to be substantially more powerful than classical methods based on nearest-neighbor matching, especially in the presence of irrelevant covariates.

TL;DR

Abstract

Authors

References92 items

Generalized random forests

Recursive Partitioning and Regression Trees

Statistical learning and selective inference

Post hoc subgroups in clinical trials: Anathema or analytics?

Prediction Policy Problems.

Recursive partitioning for heterogeneous causal effects

Uniform Convergence of Random Forests via Adaptive Concentration

Adaptive Concentration of Regression Trees, with Application to Random Forests

Robust Tree-based Causal Inference for Complex Ad Effectiveness Analysis

Valid Post-Selection and Post-Regularization Inference: An Elementary, General Approach

Heterogeneous Treatment Effects in Digital Experimentation

A Nonparametric Bayesian Analysis of Heterogenous Treatment Effects in Digital Experimentation

Estimation and Accuracy After Model Selection

Consistency of Random Forests

Asymptotic Theory for Random Forests

Quantifying Uncertainty in Random Forests via Confidence Intervals and Hypothesis Tests

Ensemble Trees and CLTs: Statistical Inference for Supervised Learning

Confidence intervals for random forests: the jackknife and the infinitesimal jackknife

Narrowing the Gap: Random Forests In Theory and In Practice

Assessing lack of common support in causal inference using bayesian nonparametrics: Implications for evaluating the effect of breastfeeding on children's cognitive outcomes

Valid post-selection inference

Estimating treatment effect heterogeneity in randomized program evaluation

From concepts, theory, and evidence of heterogeneity of treatment effects to methodological approaches: a primer

A Simple Method for Detecting Interactions between a Treatment and a Large Number of Covariates

Modeling Heterogeneous Treatment Effects in Survey Experiments with Bayesian Additive Regression Trees

Subgroup identification from randomized clinical trial data

Bootstrap-Based Variance Estimators for A Bagging Predictor.

Doubly Robust Policy Evaluation and Learning

Optimal weighted nearest neighbour classifiers

Bayesian Nonparametric Modeling for Causal Inference

On the layered nearest neighbour estimate, the bagged nearest neighbour estimate and the random forest method in regression and classification

Propensity score estimation: neural networks, support vector machines, decision trees (CART), and meta-classifiers as alternatives to logistic regression.

Analysis of a Random Forests Model

On the Rate of Convergence of the Bagged Nearest Neighbor Estimate

Subgroup Analysis via Recursive Partitioning

Asymptotics for Statistical Treatment Rules

Non‐parametric tests for distributional treatment effect for randomly censored responses

The offset tree for learning with partial labels

Inferring Welfare Maximizing Treatment Assignment Under Budget Constraints

BART: Bayesian Additive Regression Trees

Model-Based Recursive Partitioning

Consistency of Random Forests and Other Averaging Classifiers

On bagging and nonlinear estimation

Bias in random forest variable importance measures: Illustrations, sources and a solution

Quantile Regression Forests

Random Forests and Adaptive Nearest Neighbors

Nonparametric Tests for Treatment Effect Heterogeneity

OBSERVATIONS ON BAGGING

Subgroup analysis in clinical trials

Statistical treatment rules for heterogeneous populations

Statistical Modeling: The Two Cultures (with comments and a rejoinder by the author)

Subsampling

Subgroup analysis and other (mis)uses of baseline data in clinical trials

Efficient Estimation of Average Treatment Effects Using the Estimated Propensity Score

Some Comments on Cp

Bagging Predictors

Subgroup analysis of clinical trials.

SUBGROUP ANALYSIS

On Asymptotically Efficient Estimation in Semiparametric Models

Statistics and Causal Inference

The central role of the propensity score in observational studies for causal effects

The Jackknife Estimate of Variance

Probability and Measure

Estimating causal effects of treatments in randomized and nonrandomized studies.

Some Comments onCp

Asymptotic Normality of Simple Linear Rank Statistics Under Alternatives II

A Class of Statistics with Asymptotically Normal Distribution

Asymptotic statistics

Bayesian group factor analysis with structured sparsity

Causal Inference in Statistics, Social, and Biomedical Sciences

rpart: Recursive Partitioning and Regression Trees, 2014. URL http://CRAN.R-project.org/package=rpart. R package version 4.1-8

University of North Carolina at Chapel Hill

FNN: Fast Nearest Neighbor Search

Standard errors for bagged and random forest estimators

Classification and Regression by randomForest

Identifying Informative Biological Markers in HighDimensional Genomic Data and Clinical Trials