Nonparametric Goodness-of-fit Testing under Covariate Shift
Preprint on arXiv, 2026
Recommended citation: Hou, Z., & Xia, D. (2026). "Nonparametric Goodness-of-fit Testing under Covariate Shift." arXiv preprint arXiv:2608.04860. https://arxiv.org/abs/2608.04860
Authors: Zhen Hou and Dong Xia
This paper studies nonparametric goodness-of-fit tests when labelled observations come from a source population and the regression model is assessed in a different target population. It combines kernel ridge regression with truncated importance weights and a multiplier bootstrap to build confidence sets, including when density ratios have heavy tails. The paper provides finite-sample coverage guarantees and numerical experiments.
| arXiv |