Multigene Genetic Programming Based Fuzzy Regression for Modelling Customer Satisfaction Based on Online Reviews

Abstract

As markets become increasingly competitive, most businesses have adopted modern practices that helps them to enhance the competitiveness of their products. Such practices involve the use of internet though which companies gain insights into the concerns of their customers. For instance, the proliferation of e-commerce websites has enabled consumers to voice their opinions on the products they have purchased. This study proposes a methodology for modelling customer satisfaction (CS) based on online reviews using a new multigene genetic programming based fuzzy regression (MGGP-FR). Polynomial structures of CS models were developed by employing the multigene genetic programming method. The fuzzy coefficients of the polynomial structures were then determined using the fuzzy regression analysis. The proposed method was illustrated using an electronic hair dryer as a case study. The validation test results indicated that MGGP-FR the outperformed the genetic programming based fuzzy regression and the fuzzy regression analysis in terms of prediction errors.