Premium
Highly Resistant Regression and Object Matching
Author(s) -
Dryden Ian L.,
Walker Gary
Publication year - 1999
Publication title -
biometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.298
H-Index - 130
eISSN - 1541-0420
pISSN - 0006-341X
DOI - 10.1111/j.0006-341x.1999.00820.x
Subject(s) - matching (statistics) , regression , computer science , object (grammar) , artificial intelligence , statistics , mathematics
Summary. In many disciplines, it is of great importance to match objects. Procrustes analysis is a popular method for comparing labeled point configurations based on a least squares criterion. We consider alternative procedures that are highly resistant t o outlier points, and we describe an application in electrophoretic gel matching. We consider procedures based on S estimators, least median of squares, and least quartile difference estimators. Practical implementation issues are discussed, including random subset selection and intelligent subset selection (where subsets with small size or near collinear subsets are ignored). The relative performances of the resistant and Procrustes methods are examined in a simulation study.