causalTransportR : a package for bridging causal effects to new or general populations

Documentation website

causalTransportR implements a number of estimators to generalize and transport causal effects by reweighting doubly-robust score functions with transformations of selection scores. All nuisance functions are cross-fit using fast supervised learning algorithms.

Estimators

The ateGT function implements the following estimators by aggregating estimates of individual marginal means over different marginal $X$ distributions for the generalization and transportation case. $\mu, \pi, \rho$ are nuisance parameters fit using ML.

Installation

# install.packages("remotes") # if remotes isn't installed
remotes::install_github("Netflix-Skunkworks/causalTransportR")

Reference

Lal, Apoorva, Wenjing Zheng, Simon Ejdemyr, "A Framework for Generalization and Transportation of Causal Estimates under Covariate Shift", 2022