Talk by Dr. Jennifer Hill (Professor of Applied Statistics and Data Science, NYU)

4/26/2019   Wilson Hall 1-134   4:10-5:00pm

Abstract: There has been increasing interest in the past decade in use of machine learning tools in causal inference to help reduce reliance on parametric assumptions and allow for more accurate estimation of heterogeneous effects.  This talk reviews the work in this area that capitalizes on Bayesian Additive Regression Trees, an algorithm that embeds a tree-based machine learning technique within a Bayesian framework to allow for flexible estimation and valid assessments of uncertainty.  It will further describe extensions of the original work to address common issues in causal inference:  lack of common support, violations of the ignorability assumption, and generalizability of results to broader populations.  It will also describe existing R packages for traditional BART implementation as well as debut a new R package for causal inference using BART, bartCause.