Interpreting and forecasting student grades in today's higher education institutions have become a difficult task. Predicting and assessing student achievement is a critical task at institutions like universities, colleges, and school. The main aim is to concentrate on developing an improved algorithm and application to understand student performance prediction in educational system. The performance measure can be improved by using several classifiers and comparing them to previous research to investigate student performance prediction. The models are more precise than traditional models. The proposed work is evaluated using student academic performance which is taken from Kaggle website. The information gain and entropy values are used to choose the required characteristics from all of the features. The technique used for selecting the features is correlation coefficient which is one among the filter method. The models used are Naïve Bayes, XG Boost, Decision Tree and Hybrid model. Hybrid model is the combination of XG Boost and Random Forest algorithms. An accuracy of 98.39%, 96.42%, 86.55% and 74.13% is obtained for Hybrid Model, XG Boost, Decision Tree and Naïve Bayes. Among all the models the better accuracy is obtained for Hybrid model.

SMOTE, Decision Tree, Naïve Bayes