Site Search

Colloquium - Dept. Mathematics and Statistics presents Dr. S. Ejaz Ahmed, Dean, Faculty of Mathematics & Science

Add this event into your calendar using the iCAL format
  • Fri, 10/09/2015 - 3:00pm

Title:  BIG DATA: Ideas, Tips and Trick


In this talk, I will give some historical developments in the arena of so-called big data analysis.  I will shed some lights on the use and abuse of statistical techniques when analyzing such data.  I will consider  high-dimensional settings where number of variables is greater than   observations, or when number of variables are increasing with the sample size. In this context, many penalized regularization strategies were studied for simultaneous variable selection and post-estimation.  Penalty estimation strategy yields good results when the model at is assumed to be sparse.  However, in a real scenario a model may include both sparse signals and weak  signals.  In this setting variable selection methods may not distinguish predictors with weak  signals and sparse signals and may treat weak signals as sparse signals. The prediction based on a selected submodel may not be preferable due to selection bias. We suggest a high-dimensional shrinkage estimation strategy to improve the prediction performance of a submodel. The relative performance of the proposed strategy is      investigated. Various numerical studies are carried out to illustrate the superiority of the proposed strategy and an application to a microarray data is presented . 

Dina Labelle
519-253-3000 ext.3016