A Method for Detecting Quality Elements in the Kano Model based on SHAP Values

The paper proposes a method for detecting quality attributes that affect customer satisfaction from questionnaire data. By applying SHAP to the analysis of customer satisfaction survey data, we made it possible to semi-automatically classify quality elements based on the Kano model. In the proposed method, a Kano model is constructed through knowledge extraction from a learning model using questionnaire data as training data. As a result, existing customer satisfaction survey data can be utilized without investigating the degree of dependency or non-dependence of each quality characteristic. The proposed method offers many benefits to analysts, such as improved efficiency through effective use of data and assistance in extracting knowledge from data. In this study, the proposed method was applied to large-scale questionnaire data from the Oricon Customer Satisfaction Survey to demonstrate its effectiveness.