Partitioned fuzzy integral nested logit model for the Taiwan's internet telephony market

Abstract

The multinomial logit (MNL) model is the most widely used discrete choice model. It assumes theindependence of irrelevant alternatives (IIA) and the independence of attributes. The nested logit (NL) model has proven to be the most successful in solving IIA problem. However, the attributes of the NL model remain independent while attributes of social problems are often interdependent. Some researchers suggest using a fuzzy integral, and applying the idea of a fuzzy measure to solve the non-additive problem. Yet this method is unsuitable for stated preference data, which is based on hypothetical choice data. Factor analysis, the Choquet integral, and the NL model combine in this study using the partitioned fuzzy integral nested logit (PFINL) model as an innovative way to solve these problems. Taiwan's Internet Telephony market illustrates this model, and shows that the PFINL model is superior to the MNL and the NL model.

Keywords:

• internet telephony
• fuzzy integral
• multinomial logit model
• nested logit model