Online Loan Risk Assessment of College Students' Consumption Behavior : Based on Information Entropy and High-Risk Behavior

This study is dedicated to exploring a new risk assessment method for college students' consumption behavior, which cleverly integrates the information entropy theory with the analysis of high-risk behaviors of online loans, aiming to provide a solid scientific basis for the prevention of online loan risks in colleges. Firstly, the consumer behavior data of college students are collected by designing and distributing questionnaires. Secondly, based on information entropy and high-risk online loan behaviors, key risk features are extracted from the collected consumer behavior data. Thirdly, based on the extracted consumer behavior risk features, the sample data is labeled with consumer behavior risk comparison labels by expert voting. Finally, the consumer behavior risk assessment models are constructed using random forest, K nearest neighbor, and support vector machine algorithms, subsequently compared and verified with risk assessment labels to evaluate their effectiveness and accuracy. The results show that support vector machine-based consumer behavior risk assessment model achieves the highest overall identification accuracy of 85.71%, objectively and accurately revealing consumption level fluctuations and risk characteristics, offering strong support for preventing college online loan risks and elevating the quality of education, with significant innovative and practical value.