Simulation Studies for Statistical Procedures: Why Can't We Practice What We Preach?

Webinar Presented By:  Professor Hugh Chipman, Mathematics and Statistics, Acadia University

Speaker Bio:  Hugh is involved in research on a variety of modern, computationally intensive statistical methods, especially Bayesian methods for supervised learning.  These include decision trees, ensemble models, and variable selection in regression.  He is also interested in clustering, functional data analysis, and the analysis of social network data.  He has consulted with a variety of businesses, industries and government agencies over the last decade.  Prior to joining Acadia in 2004, he held faculty positions at the University of Chicago and the University of Waterloo.

Abstract:  When a new statistic, model, or learning algorithm is developed, simulation studies are often used to examine sampling behaviour or other performance measures. These studies are experimental designs. Yet the most common designs are, to put it politely, unsophisticated: either varying a single factor at a time, or a time-consuming full-factorial study. On the analysis side, basic tools like ANOVA are often forgotten. Instead, voluminous tables are presented! We demonstrate simple concepts from the design and analysis of experiments. Such tools can do for statistical simulations what statisticians promise to other scientists: save time and make efficient use of data.

Date and Time:  April 7, 2021 | 3:00 p.m Eastern Time (US and Canada)

Location:  Online

Free Event | All Welcome | Register Here

**Registration is required in order to receive the Zoom link and passcode to attend this event.  Once you are registered a confirmation email will be sent to you containing information about joining the meeting.**