An ANN-SHAP-Based Approach for Public Concerns Area Analysis in Public Service Satisfaction Surveys

The provision of public services is a core responsibility of modern governments, and optimizing this provision requires accurately identifying public concerns. Satisfaction surveys are commonly used to gather public opinion. By analyzing survey data and identifying key indicators, governments can pinpoint areas of public concern. However, traditional structural equation modeling struggles with capturing nonlinear relationships in data. This study addresses this limitation by proposing an analysis approach based on artificial neural networks and the Shapley Additivity method. Using data from two public service satisfaction surveys conducted in Hubei Province in 2023, the approach analyzes public satisfaction and loyalty in 12 service areas. Results show that the proposed model meets accuracy requirements for data analysis in both surveys. In the first survey, key concerns were government services, public utilities, and public education, while in the second, the focus was on public utilities, public education, and public employment.