Interpreting Student Performance through Predictive Learning Analytics

Abstract

In today's information-rich world, accurately predicting student performance is crucial for institutions seeking to support at-risk students and ensure their success, but this task can be challenging. Learning analytics (LA) can help identify students who are struggling and provide them with the tools and opportunities they need to succeed, benefiting both students and institutions. However, data integration from various sources can be challenging in learning analytics, causing educators to struggle with managing and keeping track of students' progress and dropouts. The goal of this project is to generate insights into student performance through the application of machine learning methods, including Random Forest (RF), Artificial Neural Network (ANN), and Support Vector Machine (SVM). These methods were used to predict students' future results and the likelihood of students' dropout based on predictive learning analytics. RF, ANN, and SVM predictive models were constructed to predict students' future final results and dropout. The dataset from The Open University is used in this study, which consists of information from multiple aspects including data about courses, student registration, results, and their interactions with virtual learning environment. RF, ANN, and SVM models were constructed to predict students' future final results based on their learning behaviour. The performance of the models is evaluated based on accuracy, precision, recall, and time taken for training. In this study, the RF model demonstrated the best performance among the three predictive models for predicting final results and dropout with the shortest training time and achieved the highest accuracy. The RF model achieved an accuracy of 87.8% in predicting final results and 82.3% in predicting dropout while maintaining an average training time of 3.6 seconds. At the end of the study, the dashboards visually presented the results, offering valuable insights into students' learning outcomes. This enables educators to effectively support their students by utilizing predictive analytics, which includes identifying potential dropouts and tailoring assistance based on these predictions.