Students Academic Performance Prediction Using an Ensemble of Machine Learning Models

Abstract

Instructional statistics mining (EDM) is an emerging subject because of the growth of tutorial records. While fact-mining techniques are used in training, they can discover hidden expertise and patterns that could aid in selection-making tactics to enhance the academic gadget. The techniques extracted from academic statistics Mining subjects are then used to apprehend college students, including their studying behavior and expecting their educational performance. This research used an EDM method to classify and predict overall performance using machine studying (ML) techniques. Pre-processing is achieved using data formatting but lacks records management, facts normalization, and facts filtering. After pre-processing, it is going to be shipped to the classifier for pupil academic performance prediction by the use of ML models like aid vector machine (SVM), Logistic regression (LR), Adaboost, selection tree (DT), and Random woodland (RF). With ML fashions, we use parameter optimization strategies like gradient descent to enhance accuracy. An experimental result demonstrates the efficiency of the projectedclassical in phrases of accuracy. The model Adaboost suggests satisfactory overall performance by achieving an accuracy of 70.39%.