A0477
Title: Parsimonious model averaging for high-dimensional data
Authors: Xinyu Zhang - Academy of Mathematics and Systems Science, Chinese Academy of Sciences (China) [presenting]
Abstract: Model averaging generally provides better prediction than model selection, but the existing model averaging methods cannot lead to parsimonious models. Parsimony is an especially important property when modeling high-dimensional data. Studying model averaging for high-dimensional data, we suggest a criterion for choosing weights. The resulting model averaging estimators of coefficients have many zeros and thus leads to a parsimonious model. The asymptotic distribution of the estimators is also provided. Furthermore, the proposed procedure is asymptotically optimal in the sense that its squared loss and risk are asymptotically identical to those of the best but infeasible model averaging estimator. Numerical analysis in comparison with the existing model averaging and selection methods strongly favors our new model averaging procedure.
