Data Integration in High Dimension With Multiple Quantiles

This article deals with the analysis of high dimensional data that come from multiple sources (experiments) and thus have different possibly correlated responses, but share same set predictors. The measurements predictors may be across experiments. We introduce a new regression approach quantiles to select those affect any responses at quantile level estimate nonzero parameters. Our estimator is minimizer penalized objective function, which aggregates establish model selection consistency asymptotic normality estimator. In addition we present an information criterion, can also used for consistent selection. Simulations two applications illustrate advantages our method, takes group structure induced by experiments levels into account.