Toward a New Hybrid System for the Prediction of Student Academic Performance

Résumé

Accurately predicting Student Academic Performance (SAP) is a complex task due to the interplay of cognitive, behavioral, and institutional factors. This paper presents a method for predicting SAP using a hybrid classification algorithm combining four complementary algorithms: Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Networks (ANNs). We employ a dataset of 32,000 students from the Open University of the United Kingdom for training and testing our hybrid system. The dataset contains information on student registration, courses, assessments, and interactions with resources. The algorithm is implemented using MATLAB software. We evaluate our system on several standard datasets and compare its performance with that of individual algorithms. The proposed method achieves an accuracy of 95.35%, a precision of 97.62%, a recall of 96.33%, and an F1-score of 96.97%. We used McNemar's test to assess the statistical significance of the model's performance, which reveals a very low p-value, indicating the robustness of the results. This system will help teachers predict students' exam results and take proactive measures to prevent academic failure.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0