Predicting Students Performance in Moodle Platforms Using Machine Learning Algorithms

E-learning platforms have become a widely used and advanced media to enhance the educational process. They bring benefits to all participants in this process – teachers, students and administration – in several different areas like teaching, learning, communication and sharing. The paper focuses on the appli-cation of machine learning algorithms for predicting students’ performance based on their interaction with the e-learning platforms. The research hypothesis is that the success or failure on e-learn courses could be predicted using data from activity logs. To support the hypothesis several machine learning algo-rithms have been performed, such as logistic regression, random forest, gradient boosting decision trees (xgboost) and neural network. The results indicate that all algorithms perform the classification task satisfactory with accuracy above 0.84. The comparison of the evaluation metrics reveals a better performance for neural network and gradient descent boosting trees compared to logistic regres-sion and random forest. The experiments have been performed using R programming language.