Inference for sparse linear regression based on the leave-one-covariate-out solution path

Abstract

We propose a new measure of variable importance in high-dimensional regression based on the change in the LASSO solution path when one covariate is left out. The proposed procedure provides a novel way to calculate variable importance and conduct variable screening. In addition, our procedure allows for the construction of p-values for testing whether each coefficient is equal to zero as well as for testing hypotheses involving multiple regression coefficients simultaneously; bootstrap techniques are used to construct the null distribution. For low-dimensional linear models, our method can achieve higher power than the t-test. Extensive simulations are provided to show the effectiveness of our method. In the high-dimensional setting, our proposed solution path based test achieves greater power than some other recently developed high-dimensional inference methods. We extend our method to logistic regression and demonstrate in simulation that our leave-one-covariate-out solution path tests can provide accurate p-values.

Keywords:

• High-dimensional inference
• regression
• variable selection
• bootstrap

Additional information

Funding

This work was partially supported by Grant R03 AI135614 from the National Institutes of Health.