Stability of a clustering structure: Client segments of a cultural institution

• Autor(a) principal: Cardoso, Margarid G. M. S.
• Data de Publicação: 2018
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://doi.org/10.34627/rcc.v5i0.27

This work implements, as a means of assessing the stability of a cluster, a new proposal for cross-validation of clusters that dispenses with the use of classifiers, resorting to the use of weighted samples of training and testing (Cardoso, Facel et al. 2009) We illustrate the proposed approach over a cluster of clients of CCB – Cultura Centre of Belem (Centro Cultural de Belém). The clustering is obtained by means of an estimation of a mixture finite model. In the constitution of the clusters or segments, it it taken in consideration the ordinal nature of the clustering base variables (measurements in Likert scale) in lieu of the usual modeling that would consider the same variables as metric. In addition, we point out to some methodologies that are considered more adequate to interpret and discriminate the segments obtained.