HYBRID MODEL FOR IMPROVING STUDENT ACADEMIC PERFORMANCE

Improving student performance in the educational institutes is a major challenge. Accurate Student Performance prediction method is used as the Hybrid method in the proposed research work. The existing student performance prediction methods namely Neural Networks, Support Vector Machine and Random Forest achieved the lower efficiency and data imbalance problem. In this research, the Hybrid Model (HM) is proposed in order to enhance student related academic accomplishment of performance prediction. The Hybrid model is based on Grey Wolf Optimization (GWO) algorithm; Synthetic Minority Oversampling Technique (SMOTE) sampling and Random Forest Classifier. The GWO method selects the relevant features from the datasets and SMOTE solve the data imbalance problem. SMOTE method overcome the imbalance problem by oversample the positive instance based on its neighbour value. The Random Forest selects the important features for the students for enhancing the performance in academics. The Hybrid model has the accuracy of 98.8% in UCI Portuguese dataset, while existing method has 95.38% accuracy.