A clustering algorithm for ipsative variables
Jessica Rubiano-Moreno, Carlos Alonso Malaver, Samuel Nucamendi‐Guillén, Carlos Eduardo López-Hernández · DYNA · 2019
The aim of this study is to introduce a new clustering method for ipsatives variables. This method can be used for nominals or ordinals variables for which responses must be mutually exclusive, and it is independent of data distribution. The proposed method is applied to outline motivational profiles for individuals based on a declared preferences set. A case study is used to analyze the performance of the proposed algorithm by comparing proposed method results versus the PAM method. Results show that proposed method generate a better segmentation and differentiated groups. An extensive study was conducted to validate the performance clustering method against a set of random groups by clustering measures.