What is the probability of selecting the best t of k populations when k is extremely large?

Start: 01/27/2011 - 4:15pm
End  : 01/27/2011 - 5:15pm

Statistics/OR/Math Finance Seminar

Jason Wilson, Biola University


In the 1950's Robert Bechhofer and Shanti S. Gupta laid the foundations of what has become known as Ranking and Selection Methods (RSM). The methods are useful for situations in which it is desired to select from among k populations, as opposed to test for significance. The area is highly developed for most of its problems when k is small (e.g. k < 20). In this talk, we will explore the issue of selection when k becomes extremely large (e.g. >1000). In particular, we will exhibit two new selection rules that are suitable for such high dimensional problems. The focus of the talk will be on on calculating the probability that such selections are correct (PCS). Under realistic conditions, older selection rules generally have PCS go to zero as k becomes large. However, under our new rules useful probabilities can be found. Some results of a simulation study will be presented which demonstrate the accuracy of estimating PCS. An example using a real microarray data set will be shown.

Harvey Mudd College 3rd floor Sprague. Refreshments at 4pm.

Claremont Graduate University | Claremont McKenna | Harvey Mudd | Pitzer | Pomona | Scripps
Proudly Serving Math Community at the Claremont Colleges Since 2007
Copyright © 2018 Claremont Center for the Mathematical Sciences