Research And Application Of Data Mining In Campus Card Consumption Behavior Analysis

With the continuously deepening reform of the informalization in colleges and universities,the university has a higher demand for the data accumulated by informalization. The global information platform such as campus card and shared databases provides a data base for data mining. This paper analyzes deeply the data mining technology, applies it to campus card consumption analysis and studies the issue of classification and association rules mining which has important significance in understanding students'consuming behavior and decision analysis.The paper studies data classification technology in data mining, focuses on the analysis of support vector machine method and Apriori association rules mining method and applies them to campus card consumption analysis.First of all, for the large number and multi-dimensional features of campus card data, the paper chooses support vector machine method with RBF as the kernel function and achieves the non-linear multi-classification of campus card consumption.Secondly, when using association rules to extract support vector machine classification frequent pattern, the efficiency is not high for multiple scan, so the paper proposes association rules improvement ideas of combining Boolean matrix and incremental approach. The method of using matrix computing features and incremental approach enhances the efficiency of Apriori mining relatively effectively.Finally, for the application background of campus card consumption, the paper provides the overall mining framework model of campus card consumption classification and association rules. It also proposes the feedback and optimization of the model and achieves the classification and association analysis mining of campus card consumption. Moreover, the paper carries out the mining experiment of the actual campus card consumption. The results show that sample selection and kernel function parameters play a key role in the classification results when using support vector machine classification. The experiment results verify the correctness of the ideas of using association rules high-frequent items to feedback support vector machine optimal sample. At the same time, the results verify the feasibility of using Boolean matrix and incremental approach to enhance the efficiency of Apriori association rules algorithm.