Zudi Lu is Professor within Mathematical Sciences at the University of Southampton. Prof Zudi LU joined, as a Professor/Chair in Statistics, in Mathematical Sciences Academic Unit and Southampton Statistical Sciences Research Institute (S3RI) at University of Southampton, UK, in late 2013. Prior to that, he had worked at several international academic institutions, including the University of Adelaide (2009-2013) and Curtin University (2006-2009) in Australia, the London School of Economics (2003-2006) in the UK, the Academy of Mathematics and Systems Science (1997-2003) in Beijing, China, and the Universite Catholique de Louvain (1996-1997) in Louvain-la-Neuve, Belgium, after he received his PhD degree from the Chinese Academy of Sciences in 1996. He was a recipient of the Australian Research Council Future Fellowship in its 2010 round, and is an elected member of the International Statistical Institute.

TITLE OF TALK: Semiparametric Model Averaging for Dynamic Time Series Forecasting: Methodology and Application

ABSTRACT: In this talk I will review some recent progress on semiparametric model averaging schemes for nonlinear dynamic time series regression models with a very large number of covariates including exogenous regressors and autoregressive lags. Our objective is to obtain more accurate estimates and forecasts of time series by using a large number of conditioning information variables in a nonparametric way. We (my coauthors including Jia Chen, Degui Li and Oliver Linton) have proposed several semiparametric penalized methods of Model Averaging MArginal Regression (MAMAR) for the regressors and autoregressors either through an initial screening procedure to screen out the regressors whose marginal contributions are not significant in estimating the joint multivariate regression function or by imposing an approximate factor modelling structure on the ultrahigh dimensional exogenous regressors with principal component analysis used to estimate the latent common factors. In either case, we construct the optimal combination of the significant marginal regression and autoregression functions to approximate the objective joint multivariate regression function. Asymptotic properties for these schemes are derived under some regularity conditions. Empirical applications of the proposed methodology to forecasting the economic risk, such as inflation risk in the UK, will be demonstrated.