An integrated machine learning and DEMATEL approach for feature preference and purchase intention modelling

Abstract

This article models the purchasing intention of young Indian consumers for branded smartphones by analysing their individual preferences for the different features of the smartphone. This article selects thirteen features for analysis, and the respondents are asked to rate, which are further used to model the mentioned purchasing intention. The features are selected using three quantitative methods: uncertainty calculation, correlation and Decision-Making Trial and Evaluation Laboratory (DEMATEL). The uncertainty indicates the variance in the respondents’ preference for a particular feature. So, the features with highly uncertain preference values are undesired. The correlation analysis indicates the feature preferences impacting the buying intention of a branded smartphone. The hierarchy and the causality for the impact of feature preferences on the branded smartphone preference are identified based on DEMATEL. The selected features are used to model the purchasing intention based on K nearest neighbour (KNN), a machine learning technique. The result is compared with the other three machine learning techniques for classification, namely, C4.5, CART, and logistic regression. The result indicates the superiority of KNN based model, which is further used for building a brand switching intention model.