Advances and Applications in Statistics
Volume 12, Issue 1, Pages 85 - 98
(June 2009)
|
|
ADAPTIVE NON-LINEAR PRINCIPAL COMPONENT AND SURFACE ANALYSIS
R. Lombardo (Italy), G. Tessitore (Italy) and J. L. A. VAN Rijckevorsel (The Netherlands)
|
Abstract: This paper suggests a method to generalise principal component analysis by transforming in adaptive way both original variables and their principal components by non-linear functions, i.e., B-splines. The main advantage is improving the model fit, taking into account non-linear relationships, and displaying data in a clearer way via principal surfaces. |
Keywords and phrases: nonlinear principal component analysis, principal surfaces, B-spline approximation, bivariate adaptive regression splines. |
|
Number of Downloads: 359 | Number of Views: 1187 |
|