A0301
Title: Factor-adjusted model averaging
Authors: Wenhui Li - Academy of Mathematics and Systems Science, Chinese Academy of Sciences (China) [presenting]
Xinyu Zhang - Academy of Mathematics and Systems Science, Chinese Academy of Sciences (China)
Abstract: A model averaging method is proposed for high-dimensional regression with highly correlated covariates. A factor structure is used to model the covariate dependence, allowing the covariates to be decomposed into two uncorrelated or weakly correlated latent components: Common factors and idiosyncratic components. The number of common factors is allowed to diverge. Estimators are averaged from factor-adjusted candidate models with augmented predictors composed of estimated common factors and idiosyncratic components. The asymptotic optimality is proven in the sense of achieving the lowest squared loss and consistency when correctly specified models exist in the model space. Numerical experiments and a real-data analysis illustrate the promising performance of the proposed method.
