Archetypal Analysis: A New Approach to Classifying Researchers

Alongside high-level publications, rankings are assuming an ever growing importance in the academic world. This article presents a new process for classifying academics, the archetypal analysis, and illustrates how it works using data from the RePEc network, one of the most comprehensive databanks on the academic output generated by economic sciences. Archetypal analysis represents a new approach to classification, which enables the identification of extreme observations in a high-dimensional dataset. These extreme values are described as archetypes. Building on these descriptions, all of the researchers covered by the data set can be allocated archetypes pro rata. The approach enables a differentiated representation of characterisations that can exceed the number of variables used.