|
Activity Number:
|
562
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 6, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Learning and Data Mining
|
| Abstract - #303756 |
|
Title:
|
On Ridge Regression, Convexity, and Nonlinear Problems
|
|
Author(s):
|
Prasenjit Kapat*+ and Prem K. Goel
|
|
Companies:
|
The Ohio State University and The Ohio State University
|
|
Address:
|
1958 Neil Avenue, Cockins Hall, Room 404, Columbus, OH, 43210,
|
|
Keywords:
|
constrained optimization ; Karush-Kuhn-Tucker conditions ; ridge regression ; sufficiency of Lagrange's principle
|
|
Abstract:
|
A recent article (Jensen and Ramirez, 2008, Intl. Stat. Rev) argued that the foundations of ridge regression are flawed and the conventional ridge estimators are not the Lagrangian solutions to constrained length problems. Using real data, they showed that the L2 norm of the solution was not strictly decreasing as a function of the penalty coefficient! Probing these issues, it is found that the regression coefficients used in their example are rescaled version of the actual ridge estimators. The seeming anomalies arise due to misinterpreting the solutions and assertions of the original Hoerl and Kennard (1970, Technometrics) paper.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |
