Toward Machine Learning and Big Data Approaches for Learning Analytics

Abstract

There is a paradigm shift in education due to online learning approaches and virtual learning environments. Machine learning methods have been used in a limited manner previously for learning analytics. These models can predict learning outcomes and enable understanding relationships between various learning variables. The data required for such predictions are usually complex with multiple relationships. In this paper, we use Support Vector Regression and Graph representation on the Open University Learning Analytics Dataset to provide a view into the use of machine learning methods and graph databases in creating predictive models for bettering the learning approaches.