Abstract:

Community governance is a key component of grassroot administration management system in our country， and also serves as the “last mile” for social governance. It also plays a crucial role in maintaining social stability and promoting the harmonious development of communities. With the rapid development of society and the continuous advancement of urbanization， the complexity and diversity of grass-roots communities are increasingly prominent， putting forward higher requirements for community inspection work. Effective community inspection can timely discover and solve problems in grass-roots governance and lay a solid foundation for the stable development of communities. However， previous research related to community governance is basically limited to case studies or qualitative analytics， and rarely adopts quantitative methodologies to analyze the problem. The inspection team and the inspected community are important components in the process of community inspection. To provide managers with more objective and interpretable decision support how the attributes of inspection teams and communities influence the effectiveness of community inspections is analyzed in this paper. The historical inspection records can be regarded as the points in the multi-dimensional space， and identifying the range of attribute values where inspection issues frequently occur can be abstracted as a frequent subset search problem in a two-dimensional attribute space. Logistic regression on the historical inspection records is employed to identify critical two-dimensional attribute combinations. Based on the results， then a mixed-integer quadratic programming model is proposed to search a space where the inspection issues frequently happen， and is transformed into a mixed-integer quadratic programming model. To enhance computational efficiency and tractability， McCormick inequalities are used to convert the model into a linear programming， and the smallest frequent subset in space prone is identified to inspection issues. Thus， the rectangle regions in the two-dimensional space， which do not overlap and are identified through the search process， represent the frequent subsets of inspection issues. To identify the attribute value ranges corresponding to a certain proportion of inspection issues， the proposed optimization model aims to minimize the sum of the areas of all frequent subsets， thereby ensuring that the distribution of inspection records with inspection issues within these subsets is more dense. Based on the practical data of community inspection records， case studies are conducted on the problems of community personnel management and community assistance， and the results demonstrate the effectiveness of the proposed method. The method can be extended to other risk analysis and influencing factor analysis problems， overcoming difficulties such as insignificant regression analysis， the inability of association rules for continuous values， and the complexity of nonlinear programming. The proposed methodology and models also offers an alternative and effective decision support framework for improving the effectiveness of community inspections. This model identifies the minimal infrequent subsets where inspection issues are likely to occur. This research assists grassroots governance departments in focusing on critical issues and provides a scientific basis for decision-making in personnel allocation and resource management. It contributes to enhancing inspection efficiency and improving overall community governance. Additionally， the study offers significant theoretical support for the improvement and development of future grassroots governance systems. The proposed framework provides valuable insights and methods for exploring more scientific and efficient grassroots governance models and facilitates the construction of a more refined and efficient inspection system.

Key words: community governance, logistic regression, spatial search method, frequent subset, mixed-integer quadratic programming