Variable Selection For Multiple Functional Regression With Serial Correlated Error

发布时间：2019-11-26 作者：浏览次数：

Speaker:	陈敏	DateTime:	2019年11月27日下午4：00
Brief Introduction to Speaker: 陈敏，中国科学院，教授。		Place:	六号楼2楼报告厅
Abstract:In this paper, we study functional linear regression with autoregressive errors. We propose a combined variable selection method to identify significant variables. In our method, a group penalty function is applied to the functional explain variables. We'll show this method identify the true model asymptotically and is almost as efficient as oracle estimator. Simulation studies and some real data examples are conducted to illustrate our result