Using Ensemble Technique and Machine Learning Algorithms to Enhance Student Performance Prediction

Abstract

Measuring The moral community relies on the students’ work. Many areas have benefited from the expanded use of machine
learning algorithms, including the prediction of diseases, student performance, agricultural output, and many more. The overarching
goal of this research is to find ways to enhance the accuracy of student performance prediction by combining several
machine learning algorithms into an ensemble method. We have used the student dataset, which includes 1000 occurrences and
22 characteristics, to assess students’ performance. We used four machine learning algorithms—Decision Tree (DT), Naïve
Bayesian (NB), K-Nearest Neighbors (KNN), and Extra Tree (ET)—in this study. Then, we created a model that uses Bagging
and Boosting ensemble techniques to aggregate the findings of each base learner. To determine the most effective model, we
examined the outcomes produced by the bagging and boosting ensemble methods. Several metrics, including sensitivity, specificity,
accuracy, and f1-score, are used to evaluate the outcomes of all machine learning algorithms and ensemble approaches.
When we examine the outcomes of bagging and bagging ensemble approaches, we discover that bagging produces the best
results. Both the institution and the admissions officers may benefit from using the created model to determine which courses
students are most likely to fail in.