A Learning Analytics Tool for Predictive Modeling of Dropout and Certificate Acquisition on MOOCs for Professional Learning

Massive Open Online Courses (MOOCs) appeared as a proper way to provide lifelong learning for potential learners of both professional and academic settings. Industry leaders can benefit from these courses because they foster the professional development of their employees in their industry. Despite these benefits, these online courses continue to register a high dropout rate and a vast number of their learners do not acquire the certificate provided at the completion of the course. This article proposes a predictive modeling tool with several Machine Learning algorithms (for generating Predictive Models) and feature engineering in MOOCs data integrated to contribute research to this specific issue. The proposed tool predicts two situations: which learners are likely to leave the course (dropout) and which learners are expected to pass the course (certificate acquisition). The tool was tested in fifteen deliveries of seven MOOCs. Initial results provide interesting information, for instance, that the accuracy of predicting certificate acquisition is higher than the precision of predicting dropout for all algorithms.