Methodological Approaches to Analyzing the Influencing Factors of Sustainable Urban Community Governance in Guangxi

Abstract Urban community governance is crucial for social cohesion and sustainable development. This study integrates Structural Equation Modeling (SEM) with machine learning methods—Artificial Neural Networks (ANN), XGBoost, and LightGBM—to examine key factors influencing Harmonized Community Development (HCD) in Guangxi, China. Based on 1,216 valid survey responses, SEM shows that Community Integration Satisfaction (CI), Resident Satisfaction (RS), and Resource Inputs (RI) are the strongest predictors of HCD. Party Leadership (PL), Community Resident Self-Governance (CRS), Collaboration of Social Forces (CSF), and Community Workforce Building (CWB) indirectly affect HCD through CI, RS, and RI.Machine learning models validate these findings: CI consistently ranks highest in importance, while RS and RI also contribute significantly. ANN captures nonlinear patterns, and XGBoost and LightGBM reinforce the variable rankings. Although PL, CRS, CSF, and CWB have lower direct impact, they offer room for targeted improvement.The study highlights the value of combining SEM and machine learning to uncover complex governance dynamics. Enhancing community integration, satisfaction, and resource support emerges as vital to building inclusive and sustainable urban communities.