Research And Application Of Student Affairs Analysis System Based On Data Mining

The various systems in colleges contain a wealth of student data,but these systems are basically lacking comprehensive analysis and decision-making capabilities.In order to utilize the student affairs data and mine the hidden information to better understand the life and learning situation of college students,a student affairs analysis system based on data mining which realizes two modules of student consumption analysis and achievement analysis is proposed.The student affairs analysis system contained data layer,analysis layer,application layer and presentation layer.Firstly,on the data layer,the preprocessing technology is used to integrate multiple data sources to form a data warehouse which was suitable for data mining.Then,at the analysis layer and application layer,different data mining algorithms for different modules in the system were researched.On the student consumption analysis module,the K-means algorithm is used for cluster analysis,and the association analysis algorithm is used to extract the strong association rules on the student achievement analysis module.Finally,the data mining structure is visualized on the presentation layer.The subject focused on the frequent item set mining such as FP-growth and CANtree in the association analysis algorithm.After analyzing the characteristics of FPgrowth and CAN-tree,combined the advantages of FP-growth,an improve CAN-tree algorithm which used the runtime reconstruction technique is proposed.The number of nodes in the tree structure was reduced effectively.At the same time,after studying the frequent item set mining algorithm of tree structure,in order to solve the problems that based on the repeated traversal of both the same node and prefix path and the repeated construction of the conditional pattern tree,a tree structure improvement strategy is proposed.The improved algorithm used Array-structure and auxiliary item header table with two layers of hash structure to locate the item node,and mine the frequent item set on the auxiliary item header table directly.The experiment is verified by different types of data sets,the tree structure improvement strategy optimized the execution efficiency of the algorithm greatly.