报告题目： New Semiparametric Estimation and Forecasting of High-Dimensional Dynamic Time Series
报 告 人： 李德柜教授 （The University of York）
报告时间 ：2018年7月2日上午 10:00-11:00
报告地点：知新楼B-1238

报告摘要：
In this talk, we introduce a flexible and easy-to-implement semiparametric approach to estimate and forecast high-dimensional time series data. This is conducted by a novel technique of Model Averaging MArginal Regression (MAMAR) with the weights chosen through a two-stage semiparametric method. Both the large-sample theory and practical application of the proposed estimation and forecasting method are given in the talk. We further study a challenging case where the number of time series variables may exceed the time series length, and combine the developed MAMAR method with the shrinkage and factor modelling approaches to achieve dimension reduction and then construct feasible estimation and prediction. Finally, we discuss the application of the MAMAR approach in estimating the large dynamic covariance matrix. This talk is based on some joint research projects with J. Chen(Economics, York), O. Linton (Economics, Cambridge) and Z. Lu (Statistics, Southampton).

欢迎各位老师同学积极参加！