JSM Activity #2001-01C


Back to main JSM 2001 Program page





Activity ID:  2001-01C
Title Room
Model-Based Clustering M-International Salon A
Date / Time Sponsor Type
08/05/2001    8:00 AM  -  4:00 PM ASA Other
Organizer: n/a
Chair: n/a
Discussant:  
CE Presenter Adrian Raftery
Description

Clustering and classification problems are prevasive in the physical, biological and social sciences as well as in engineering. Leading applications have included market segmentation and biological taxonomy, and areas of recent interest include clustering problems in the analysis of DNA microarray gene expression data, text categorization for the Web, automatic image segmentation, and datamining. The goal is to divide data into groups whose members have more in common with each other than with members of other groups. This course describes in detail a general framework for clustering based on mixture models that provides a principled statistical approach to important practical issues that arise in cluster analysis, such as determining the number of groups in the data, selecting an appropriate statistical model, and handling outliers. We show how this methodology can be applied in various clustering applications, as well as to multivariate density estimation and discriminate analysis (supervised classification). Many of these ideas have been implemented in the MCLUST software, whose development has been sponsored over a number of years by the Office of Naval Research. The course will make extensive use of examples demonstrating the use of MCLUST, which interfaces to Splus. Applications in data mining will also be discussed. Fees: M-$450 (after July 13 $575) NM-$550 (after July 13-$675) SM-$280(no discount after July 13) Continuing Education Units: 1.20
JSM 2001

For information, contact meetings@amstat.org or phone (703) 684-1221.

If you have questions about the Continuing Education program, please contact the Education Department.

Revised March 2001