時(shí)間：3.23（星期一）下午3：00

地點(diǎn)：研究生樓103

主講人：Ejaz Ahmed

摘要：In big data analysis it is commonly assumed that the predictive model is sparse. However, a model at hand may have sparse signals as well as with a number of weak signals. In this scenario aggressive variable selection procedures may not clearly distinguish predictors with weak signals and sparse signals. The prediction based on a selected submodel may not be efficient in such cases. To overcome this shortcoming, we propose a high-dimensional shrinkage estimation strategy to improve the prediction performance of a submodel.  The relative performance of the proposed prediction strategies are appraised analytically and by simulation studies and the real data analysis.

 

主講人介紹：Ejaz Ahmed is professor of statistics and Dean of Faculty of Math and Science. Before joining Brock, he was a professor and head of dept. of Math & statistics at the University of Windsor and University of Regina. Prior to that, he had a faculty position at the University of Western Ontario. His area of expertise includes statistical inference, shrinkage estimation, and big data analysis. He has numerous published articles in scientific journals, both collaborative and methodological. Further, he has written several books, edited and co-edited several volumes and special issues of scientific journals.  He organized several workshops and many invited sessions. He made numerous invited scholarly presentations in many countries around the globe. Dr Ahmed serves on the editorial board of many statistical journals and review editor for Technometrics. He served as a Board of Director and Chairman of the Education Committee of the Statistical Society of Canada. He was a VP communication for ISBIS. Recently, he served as a member of an Evaluation Group, Discovery Grants and the Grant Selection Committee, Natural Sciences and Engineering Research Council of Canada (NSERC). He is an elected Fellow of American Statistical Association.