Selective inference in regression models with groups of variables

Abstract

We provide a general mathematical framework for selective inference with supervised model selection procedures characterized by quadratic forms in the outcome variable. Forward stepwise with groups of variables is an important special case as it allows models with categorical variables or factors. Models can be chosen by AIC, BIC, or a fixed number of steps. We provide an exact significance test for each group of variables in the selected model based on an appropriately truncated $\chi$ or $F$ distribution for the cases of known and unknown ${\sigma }^2$ respectively. An efficient software implementation is available as a package in the R statistical programming language.

16 pages, 4 figures

References in corpus (1)

• A significance test for forward stepwise model selection

Cited by in corpus (11)

• Bootstrapping and Sample Splitting For High-Dimensional, Assumption-Free Inference
• Bootstrap inference after using multiple queries for model selection
• An MCMC-free approach to post-selective inference
• More Powerful and General Selective Inference for Stepwise Feature Selection using the Homotopy Continuation Approach
• Valid and Exact Statistical Inference for Multi-dimensional Multiple Change-Points by Selective Inference
• Approximate Post-Selective Inference for Regression with the Group LASSO
• Computing Valid p-values for Image Segmentation by Selective Inference
• Valid Inference Corrected for Outlier Removal
• Scalable methods for Bayesian selective inference
• Selective inference after likelihood- or test-based model selection in linear models
• Selective Inference for Latent Block Models