ABSTRACT

This chapter provides an exposition about the 'best' method of building a geodemographic, or how they might be assessed, and explores how output geodemographic patterns can be sensitive to changes in methodological approach. It outlines some potential options that a classification builder might take when building a geodemographic classification. The chapter provides an overview of how geodemographic classification emerged as a method of describing the characteristics of areas from rich multidimensional census data. Clustering algorithms attempt to seek an optimal grouping of areas into clusters by maximising some measure of within-cluster homogeneity or between-cluster heterogeneity. Historically, managing is a large number of attributes when building geodemographic classifications was more difficult with restricted computing power. Principal component analysis (PCA) was introduced as a method of reducing attribute dimensions, and reducing the impact of correlated attributes. The changes from the national extent impact the ability to use geodemographics as a measure for comparing places.