An interactive learning analytics tool to support higher education stakeholder to more explain and interpret predictive student's failure or success

Abstract

Learning Analytics has great potential for improving student learning and pre- diction making about the probability of their success or failure. However, these machine learning algorithms generally do not explain these pre- dictions. Even if a machine learning model works well, ignoring why it makes certain decisions excludes the human touch, which in turn negatively impacts the stakeholders' trust. Also, knowing the "why" can help the teaching sta decipher the data and the reason why a particular prediction was made. In this study, we are interested in providing the educators with an interactive learning analytics tool that allows them to be actors in this 'loop' of analytics and predictions, thus having a proactive role in deciding a student's future.