RdistNN website copy
Source:WEBSITE.md
Prepared for a software entry or research-page addition on jakobjuergens.com. The text below introduces the 0.1.0 release; publish it after the release and documentation links are live.
Software entry
RdistNN — local treatment effects and nonparametric regression in R
RdistNN implements distributional nearest-neighbor estimation for conditional average treatment effects and nonparametric regression. It combines local estimates with delete-1 jackknife standard errors, tuning comparisons, and diagnostics for weights and treatment overlap. CATE nuisance predictions are supplied by the user; valid inference requires the conditions developed in the accompanying research. The initial release focuses on CATE and regression. PLQR and local IV-GMM functionality are experimental.
Source and installation · Documentation · Getting started · Accompanying paper
Release announcement
RdistNN 0.1.0 is available: an R package for conditional average treatment effects and nonparametric regression with distributional nearest-neighbor weights. The package includes jackknife standard errors, localization tuning, and weight and overlap diagnostics. The getting-started guide demonstrates both data-frame and matrix interfaces and explains the inference conditions. PLQR and local IV-GMM remain experimental.