A Novel Hybrid Approach Of Adaboostm2 Algorithm And Differential Evolution For Prediction Of Student Performance

摘要

The prediction of performance of student is a very important task for institutions of higher learning. The academic performance of studentsaids the teachers, instructors and management of institutions to identify low performing students and then more attention is given to them so as toenhance their performances. In previous studies, various elements and methods have been applied to identify and enhance the performances ofstudents. In this work, we made use of data mining classifcation techniques to improve the prediction accuracy of student performance. In recent times,classification accuracy has been enhanced thru the use of ensemble techniques and combining multiple classifiers. In this paper, we have made use ofan efficient AdaBoost ensemble technique called AdaBoostM2 and we combined it with a metaheuristic optimization algorithm known as DifferentialEvolution (DE) to produce a novel algorithm called ―ADDE‖. This new algorithm is implemented on the KalBoard 360 educational dataset and the resultsdisplays that is very efficient in reducing the weak learners and thereby increases the prediction accuracy. The new algorithm has therefore shown betterresult in reducing the computation complexity.