Regression Modeling through graphics

This nine-hour workshop presents a methodology for regression based upon graphics. It moves graphics from the periphery (diagnostics only) to the center of analysis, and is applicable to almost any regression.

Regression Modeling through Graphics begins with the central goal of reducing the dimension of the vector of predictors X, without loss of information on the conditional distribution of the response Y given X, and without requiring a model. This sufficient dimension reduction leads to sufficient summary plots, which contain all of the information on the regression that is available from the sample. Such summary plots can be useful throughout a regression study, particularly for guiding the choice of a first model and for studying residuals from a target model.

The workshop describes a new context for regression that centers on dimension reduction and sufficient summary plots, and that requires few scope-limiting constraints. Topics and techniques covered include standard residual plots, inverse response plots, 3D plots, and graphical regression. Numerical methods such as sliced inverse regression (SIR), sliced average variance estimation (SAVE), and principal Hessian directions (pHd) are also discussed.

Included are lecture, computer demonstrations, and interactive computer lab work. Software that integrates the new graphical methodology with many standard regression methods is available to those enrolled at no additional charge.

Instructor:

R. Dennis Cook is Professor of Applied Statistics at the University of Minnesota, and is author of Regression Graphics: Ideas for Studying Regressions through Graphics and co-author of Introduction to Regression Graphics, both published by Wiley. He has published numerous articles in statistics and applications journals, including many on regression graphics, nonlinear modeling, regression diagnostics, optimal experimental design, and population genetics. He has consulted extensively in the physical and biological sciences, and in law. He is a Fellow of both the American Statistical Association, and the Institute of Mathematical Statistics, and a member of the International Statistical Institute.

Audience:

Scholars and researchers who want to expand the role of graphics in their work, especially college/university instructors and graduate students in any field, and industry statisticians.

Prerequisite:

Familiarity with standard regression methodology, a first course in mathematical statistics (Master's degree level a recommended minimum), and basic knowledge of linear algebra.

Provisions:

Enrollees receive substantial handouts, graphical regression computer software (used in the workshop), and a copy of Prof. Cook's book, Regression Graphics: Ideas for Studying Regressions through Graphics (Wiley, 1998, $80 list price). Morning and afternoon refreshments provided. Break time for off-site lunch (lunch not provided).

Date:

Monday and Tuesday, May 3 - 4, 1999.

Time:

8:15 a.m. - 4:30 p.m. Enrollees will be in a computer lab session on Tuesday, either morning or afternoon.

Location:

Training Room, 3rd floor, Media Union, The University of Michigan.

Size:

Maximum 50 participants (limit ten at student fee rate), 1-2 per computer.

Fee:

$250 for members of the Ann Arbor Chapter of The American Statistical Association, and for University of Michigan affiliated faculty, staff and students; $150 for student members of the Ann Arbor Chapter of The American Statistical Association; $450 for all others. (Fee includes graphical regression software and $80 value book.) Fees can be paid by check or billed to a University of Michigan Account. Join ASA and its Ann Arbor chapter for $80 individual (or $27 student) at ASA's web site: http://www.amstat.org.

Registration:

Call CSCAR at (734) 764-7828. Enrollment is limited. Fee must be paid in full by Tuesday, April 27. Make check payable to CSCAR--University of Michigan, or give the University of Michigan Account to be billed. Send check to CSCAR, University of Michigan, 3514 Rackham Bldg., 915 E. Washington St., Ann Arbor, MI 48109-1070.