Premium
Combining bilinear modelling and ridge regression
Author(s) -
Høy Martin,
Westad Frank,
Martens Harald
Publication year - 2002
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.727
Subject(s) - overfitting , principal component regression , partial least squares regression , interpretability , regression , regression analysis , mathematics , bilinear interpolation , statistics , linear regression , variance inflation factor , ridge , polynomial regression , elastic net regularization , econometrics , multicollinearity , computer science , artificial intelligence , geology , paleontology , artificial neural network
A method is presented for making principal component regression (PCR), partial least squares regression (PLSR) and other regressions based on bilinear modelling (BLM) less sensitive to overfit. The idea is to use generalized ridge regression to calculate the Y‐loadings in order to prevent small, uncertain values of the score vectors from causing inflation of variance in the regression coefficients. Thus we combine the stabilizing power of ridge regression with the modelling power and interpretability of bilinear models. The method is intended to provide better predictive ability and improved stability for regression models. Copyright © 2002 John Wiley & Sons, Ltd.