Machine learning-based inference analysis for customer preference on e-services features

Abstract

This study first proposes a set of factors and an initial behaviours-requirement relationship model as domain knowledge. Through conducting a questionnaire based survey customer data is collected as evidences for inference of the relationships between these factors shown in the model. After creating a graphical structure, this study calculates conditional probability distribution among these factors, and then conducts inference by using the Junction-tree algorithm. A set of useful findings has been obtained for customer online shopping behaviour and requirements with motivations. These findings have potential to help businesses adopting more suitable development activities.