Prediction of Student's Academic Performance using Feedforward Neural Network Augmented with Stochastic Trainers

The academic performance of students is of great interest to tutors and decision-makers in educational institutions. The extensive use of information technology systems in education generates an enormous amount of data, which is challenging to analyze and extract valuable information. Therefore, Educational Data Mining (EDM) concept emerges to adapt Data Mining (DM) techniques to extract the hidden and valuable educational knowledge that improves the learning process. The primary purpose of this paper is to introduce an efficient student's performance prediction model. For this purpose, a feed-forward Multi-Layer Perceptron approach boosted with stochastic training algorithms is proposed. The proposed model is benchmarked and assessed using three public educational datasets gathered from UCI and Kaggle repositories. Synthetic Minority Oversampling Technique (SMOTE) is utilized to handle the imbalanced data problem. The performance of the proposed model is evaluated by a set of classifiers, namely, Support Vector Machine, Decision Trees, K-Nearest Neighbors, Logistic Regression, Linear Discriminant Analysis, and Random Forest. The comparative study revealed that the MLP achieved promising prediction quality on the majority of datasets compared to other traditional classifiers, as well as those in previous works.