11 Cluster Analysis

Abstract This chapter introduces cluster analysis algorithms for finding subgroups of objects (e.g., patients, genes) in data such that objects within a subgroup are more similar to each other than to objects in other subgroups. The workhorse of cluster analysis are the proximity measures that are used to indicate how similar or dissimilar objects are to each other. Formulae for calculating proximities (distances or similarities) are presented along with issues related to scaling and normalizing variables. Three classes of clustering are presented next – hierarchical clustering, partitioning, and ordination or scaling. Finally, some recent examples from a broad range of epidemiology and medicine are very briefly described.