yixinsun1216/crossfit: Double/Debiased Machine Learning

An implementation of double/debiased machine learning method (Chernozhukov et al.2018)to remove regularization bias and overfitting in estimating parameters of interest.This method has two critical contributions: 1. The use of cross-fitting to efficiently split the sample. 2. Identifying parameters using Neyman-orthogonal moments, which are less sensitive to errors in estimating nuisance parameters. Here, we introduce select machine learning techniques that can be applied for causal inference in this framework.

install.packages("remotes")
remotes::install_github("yixinsun1216/crossfit")