The Many Faces of Multivariate Data: Chernoff Faces

Humans have the ability to discern subtle differences in facial expressions. What if multivariate data were represented in the form of a facial expression? Would it be easier to compare data sets? That is the argument behind Chernoff faces, an idea conceived by Herman Chernoff. Traditionally, Chernoff faces are a series of 2D shapes and lines that imitate facial features i.e. a triangle for a nose, a circle for an eye etc. However, Chernoff outlined only a concept; he did not specify an algorithm to implement this idea. Given a data set, each variable would be mapped to a particular facial feature and morphed accordingly. The ultimate goal is to create a web-based application that takes in multiple data sets and generates a Chernoff face for each set in order to juxtapose the data. Project Blog: lilianasnrdes.wordpress.com Live Project: lilianamatos.com/chernoff/