Multidimensional Data Mining Model Analysis of Financial Structure Optimization of Universities in Government Accounting Comprehensive Budget Performance Management

Abstract

Under the background of government accounting comprehensive budget performance management, this study focuses on the optimization of financial structure of colleges and universities, and constructs a set of financial decision-making model system based on multi-dimensional data mining. By integrating high-precision financial data such as budget accounts, project expenditures, scientific research funds, tuition income and expenditure of universities. Innovative fusion of three types of data mining techniques. Multi-dimensional association rule mining reveals inter-departmental expenditure associations, cost composition laws, project hierarchy differences, etc. ID3 decision tree algorithm realizes financial risk classification and budget execution evaluation.DBSCAN density clustering divides student groups by the amount of outstanding fees and supports poor student classification. The data read/write efficiency of the proposed method is over 42bit/s, which is 200% higher than the benchmark method, and the redundant rules are reduced by 60%. And 415 strong association rules are generated (e.g., conference fee → postal fee, confidence level 0.6125, enhancement 1.98).The AUC of the three poverty categories classified by the DBSCAN model exceeds 0.98, which indicates that the model is more accurate and achieves a better classification effect. The decision tree model accurately quantifies the poverty class (0.247 gain in per capita annual household income of the root node), providing a data kernel for scientific allocation of resources.