Principal Component Analysis

Definition

Principal components analysis is a particular – and commonly deployed – form of factor analysis.

Principal components analysis first identifies a latent variable which is as close to (as correlated with) all of the original variables as possible. It then identifies a second latent variable uncorrelated with the first, which is as close to the residual variation in the original variables as possible; and continues until it has exhausted all of the variation, creating n latent variables, where n is the total number of original variables being studied.

Principal component analysis is widely used across the social sciences to identify general patterns in large data sets. It has been widely used, for example, in the categorization of different types of neighborhood in cities (often termed factorial ecologies) that have been used not only in descriptive analyses of the urban residential fabric but also in commercial marketing strategies – generally known as geodemographics –...