Graphical representation of multivariate data using Chernoff faces

Abstract

Chernoff (1971, Technical Report 71, Department of Statistics, Stanford University; 1973, Journal of the American Statistical Association 68: 361-368) proposed the use of cartoon-like faces to represent points in k dimensions. This article describes a Stata implementation of a face-generating algorithm using the method proposed by Flury (1980, Technical Report 3, Institute of Mathematical Statistics and Actuarial Science, Bern University), Schüpbacli (1987, Technical Report 25, Institute of Mathematical Statistics and Acturial Science, Bern University), and Friendly (1991, http://www.math.yorku.ca/SCS/ sasmac/faces.html). I present examples of applying Chernoff faces to data clustering and outlier detection. © 2009 StataCorp LP.