Research And Application Of Data Mining For Student Behavior Analysis In College

With the advance of information technology, in most universities,information technology had been used to improve the management work. Digital campus has become an important stage of information management. It has become a key problem to be solved that extracting the latent information from the accumulated large amount of data which can help university manage the students and help the students to benefit themselves. Data mining technology is of great help to solve this problem. With the data mining technology, based on the data accumulated in colleges and universities, the relationship between different behaviors such as student achievement, consumption and net charge recharging can be excavated, which can promote the building of college and improve the student management.In the paper, we describe the development of digital university and the work others had done in the analysis of university data firstly. Secondly, we briefly introduce the concept and development of data mining, and present the common techniques of data mining. After that, we describe the design of student behavior analysis system, explain the whole frame of the system and the function of each module, meanwhile the design of database is introduced. In particular, an improved K-means algorithm is proposed to cluster the data more accurately. In order to decrease the impact of outliers on K-means algorithm, we also propose the method of removing outliers based on grid density. A new initial clustering center generation method is proposed to improve the performance of K-means algorithm. Then the student behavior analysis system is implemented on the basis of design. The system is used to cluster the data of students and get the categories of students’ behaviors, while the association between the behaviors is obtained by mining association rules. The results are presented in a visualized way.Finally, we validate and verify the experimental results produced by experiments using real data. The results show that the student analysis system can effectively draw the clustering results and the association rules among the students’ behaviors. Compared with others, the method used in the system is more efficient and accurate. Therefore, the results make us have a better understand of students and provide strategic support for school management.