Research On The Technology Of University Student Achievement Warning Based On Data Mining

With a large number of college student achievement increasing, studentachievement statistics and comparison is too simple to meet the actual needs ofstudents and educators. Researching on student achievement warning system hasbecome a hot researching in the field of universities information management.This thesis uses the correlation algorithm of university student achievementinformation for data mining, Obtaining the relationship between courses to helpeducation administrators on student achievement warning. This thesis warning ofcollege student achievement related technologies following research:Apriori association rules algorithm has been improved. Apriori algorithm isanalyzed, to study the process of implementation of Apriori algorithm in practicalapplication, the analysis found: Apriori algorithm was used in actual operation lowefficiency and it was difficult to run in some special cases. To make the algorithm canmeet the actual demand, Apriori algorithm was improve to solve arithmetic problemsin actual operation by optimizing algorithm stored frequent items methods to improvethe efficiency of the algorithm, and accuracy of the algorithm. To improve theefficiency of the algorithm were compared before and after the experiment.Select the C4.5algorithm as the tree algorithm mining, and on the C4.5algorithm has been improved. Studying the process of C4.5algorithm been used inthe practical applications, the analysis found: C4.5algorithm to calculate the Studentinformation section cumbersome step in the implement process, the efficiency of thealgorithm is affected. To improve the efficiency of the algorithm, by resetting theC4.5algorithm implementation, optimization algorithm C4.5, improve the efficiencyof the algorithm.Designing the experimental system of algorithms to validate the algorithmsimproved. First, Establishing the overall architecture of the experimental system, andthen created a database for experimental system. the collected data is datapreprocessing, unified data format, loaded into the database, according to thealgorithm functions to achieve various functional modules of the system. Compare the results before and after the improved algorithm for mining association rules wasstrongly validate the feasibility and accuracy of test algorithms in practicalapplications, the improved algorithm before and after comparison of the resultsobtained were analyzed warning, testing experimental systems and algorithms theactual effectiveness of warning.