Cub model-based clustering of Likert-type data with a tourist satisfaction application

Abstract

In investigating customer satisfaction with products or services, the most popular approach still relies on interviews or questionnaires to obtain consumers’ opinions, and responses are usually measured by means of Likert-type scales. However, Likert-type data are inherently imprecise and uncertain. Thus, to obtain reliable analysis using such data, an a-posteriori correction must be adopted. The fuzzification procedure is the most common a-posteriori way to deal with uncertainty of Likert-type data. In this study, an alternative method to address the uncertainty of such data when used as input of a cluster analysis is proposed. The suggested method is based on the CUB model and the Fuzzy C-Medoids Clustering of Mixed Data algorithm and it is theoretically and empirically presented using real case study data. Advantages of the FCMd-CUB method are discussed in the conclusion section.