摘要: Publisher Summary This chapter discusses the space of Chernoff faces. Using faces to their best advantage represent multivariate data requires that perceptual configuration, in which are perceived, resemble configuration original points. Observers' judgments similarities between should reflect distances Multidimensional scaling solves this problem principle, but it is not practical because dimensionality spaces. Intuitively obvious deficiencies way face parameters map onto can be corrected by trial and error. Regions possible parameter vectors evaluated according degree subjective distance within each region match a particular model. A special case paradigm used for any specific type data.