Use of a Genetic Algorithm for Factor Selection in Principal Component Regression

Publisher Website
Google Scholar
Abstract
The critical point in the development of principal component regression (PCR) calibration programs is the automatic factor selection step. In classical methods this is based on a differentiation between primary and secondary factors and other statistical assumptions and criteria. In contrast to this the Genetic Algorithm (GA) used for factor selection in this paper finds the optimal combination of factors without statistical constraints beyond an appropriately chosen fitness function.

This publication has 3 references indexed in Scilit: