Predicting smartphone user’s satisfaction using decision tree

Abstract

Literature has reported that smartphone user’s satisfaction can differ according to consumer life experiences by using structural equation modelling (SEM). However, SEM seems too complex and difficult for the practitioners, particularly the interpretation of numerical estimates for each of the parameters, provided by SEM. Recently, the decision tree model has been used to analyze the factors affecting consumer satisfaction. Decision trees (DT) represent one of the simplest and often most effective non-parametric supervised learning method used for classification. For searching for a practical decision rule, this paper employed DT to explore decision rules used for predicting smartphone use’s satisfaction. Three constructs: consumer innovativeness, usage frequency and innovation attributes, which may influence smartphone user’s satisfaction, as well as generations as a proxy of age are considered in the model of this paper. It is found that the compatibility of innovation attribute of smartphone dominates the prediction of smartphone user’s satisfaction. Among forty-nine candidates, only two items used to measure compatibility construct were chosen as predictors of smartphone user’s satisfaction in the final model. These two items classified total respondents into three classes, with different satisfaction levels. Our results also provide a plausible explanation of two observed phenomenon: (1) why the majority of smartphone users are teenagers, and (2) why the complexity of innovation attributes is currently not important to prediction of smartphone user’s satisfaction. Finally, the managerial implications of this paper are discussed, and the direction of future research is also suggested.