Neuro-Fuzzy Systems for Learning Analytics

Learning Analytics techniques are widely used to improve students' performance. Data collected from Virtual Learning Environments (VLEs) are helpful to predict students' outcomes through the analysis of their interactions with the platform. In this work, we propose the use of hybrid models which are able to return accurate predictions together with explanations on the processes leading to the results. Particularly, students' outcomes have been predicted through Neuro-Fuzzy Systems (NFSs), and explanations are given in form of "IF-THEN" rules. The balancing between model interpretability and accuracy has been studied by reducing the number of relevant features. Results on the Open University Learning Analytics Dataset (OULAD) have shown the effectiveness of NFSs in predicting students' outcomes while explaining the reasoning behind the process.