flexBCF

flexBCF is a faster and more flexible implementation of Bayesian Causal Forests. This implementation was created as part of the 2022 American Causal Inference Competition (ACIC) Data Challenge. Please see our paper for more details.

Note that flexBCF implements a slightly different model than what is provided by default in the original bcf package (available at this link). Namely, it (i) does not use half-Normal or half-Cauchy priors for scale parameters; (ii) is not invariance to re-coding of the treatment indicator; and (iii) uses the same regression tree prior $\mu$ and $\tau$ (in contrast to bcf which more strongly regularizes $\tau$-trees towards shallow trees). If you require such functionality, please contact the package maintainer.

You can install flexBCF using

devtools::install_github(repo = "skdeshpande91/flexBCF")