Abstract
This paper explores how behavioral, academic, and parental engagement data provided within the xAPI-Edu-Data dataset can be used to predict the academic performance of students when training on machine learning models with supervised learning. Due to the developing demands of the data-driven initial selection of the learners under risk, the study will create a valid and explainable predictive model that can consider the most significant factors of student success in Learning Management System (LMS). The research is based on behavioral engagement and self-regulation learning theories; the observations included in the analysis of student interaction, i.e. resource usage, classroom engagement and engagement in discussions, in terms of their connexon to academic success. A positivist, quantitative approach was embraced with 480 student records with 16 variables of features in the category of demographic, academic, behavioral and parental involvement. Several models of supervised learning were tried and compared, among which were Random Forest, Gradient Boosting, Decision Tree, Naïve Bayes, Logistic Regression, Support Vector, and K-Nearest Neighbors. Preprocessing of data was done through categorical encoding, feature scaling where necessary and stratified train test splitting. Accuracy, precision, recall, F1-score and 10-fold cross-validation were used to assess the model performance. These findings show that the ensemble to be the most predictive in terms of accuracy (80.21 -> the highest) were the ensemble with Random Forest and a voting classifier. The strongest predictors were found to be behavioral engagement variables, with limited roles being played by the demographic factors. The results prove the efficiency of ensemble learning in educational analytics and show that behavioral information is helpful in the early intervention and evidence-based educational decision-making.
Publication Date
1-2026
Document Type
Thesis
Student Type
Graduate
Degree Name
Professional Studies (MS)
Department, Program, or Center
Graduate Programs & Research
Advisor
Khalil Al Hussaeni
Advisor/Committee Member
Sanjay Modak
Recommended Citation
Alfalasi, Saeed, "Predicting Student Academic Performance Using Behavioural and Parental Engagement Data from Learning Management Systems" (2026). Thesis. Rochester Institute of Technology. Accessed from
https://repository.rit.edu/theses/12508
Campus
RIT Dubai
