Student Academic Performance using Feature Extraction Techniques with Ensemble machine Learning Model

Abstract

Overcoming intellectual obstacles is necessary to ensure education as a human right. Some of these challenges include issues with educational systems, a lack of determination, and an unhealthy obsession with the internet. Machine learning is making rapid strides, which empowers instructors to take charge of online learning. To guarantee that students do well in school, these models have the potential to detect individuals who are susceptible to sliding below. so that they can be helped quickly. The Open University Learning Analytics (OULA) dataset is utilized in this work to identify robust grade-predicting factors. Students' past work and credit acquired are recognized as significant factors that connect to their achievements. These factors are extracted using PCA and RFE in the study. Robust prediction performance was achieved by constructing an ensemble classifier that combined the strengths of the gradient-boosting and Random Forest models. The success of this strategy is demonstrated by metrics like 87% precision, 83% recall, 88% F1 score, and 89% correctness. Results from the model also show a favorabless relationship between students' success and their prior efforts and studied credits. The method's main idea is to find at-risk students and help them right away by interpreting their predictions. Improved academic outcomes for students are a direct result of the efficient model and methods that use machine learning (ML) for educational data.