Regression modelling on stratified data with the lasso

• 成果类型: Article
• 署名作者: Ollier, E.; Viallon, V.
• 署名单位: Ecole Normale Superieure de Lyon (ENS de LYON); Universite Gustave-Eiffel; Universite Claude Bernard Lyon 1
• 刊物名称: BIOMETRIKA
• ISSN/ISSBN: 0006-3444
• DOI: 10.1093/biomet/asw065
• 发表日期: 2017
• 页码: 8396
• 关键词: GENERALIZED LINEAR-MODELS; regularization paths; effect modifiers; fused lasso; selection; RECOVERY; sparsity; RISK

摘要

We consider the estimation of regression models on strata defined using a categorical covariate, in order to identify interactions between this categorical covariate and the other predictors. A basic approach requires the choice of a reference stratum. We show that the performance of a penalized version of this approach depends on this arbitrary choice, and propose an approach that bypasses this at almost no additional computational cost. Regarding model selection consistency, our proposal mimics the strategy based on an optimal and covariate-specific choice for the reference stratum. An empirical study confirms that our proposal generally outperforms the basic approach in the identification and description of the interactions. An illustration on gene expression data is provided.