Penalized weighted composite quantile regression for partially linear varying coefficient models with missing covariates

In this paper we study partially linear varying coefficient models with missing covariates. Based on inverse probability-weighting and B-spline approximations, we propose a weighted B-spline composite quantile regression method to estimate the non-parametric function and the regression coefficients. Under some mild conditions, we establish the asymptotic normality and Horvitz–Thompson property of the proposed estimators. We further investigate a variable selection procedure by combining the proposed estimation method with adaptive LASSO. The oracle property of the proposed variable selection method is studied. Under a missing covariate scenario, two simulations with various non-normal error distributions and a real data application are conducted to assess and showcase the finite sample performance of the proposed estimation and variable selection methods.