Abstract
Four problems relating to the use of principal components analysis, based on correlation coefficients, are discussed with reference to geographical structures. The general conclusion is that the method is not suited to many geographical data sets.

This publication has 12 references indexed in Scilit: