Online application for the early detection of students at risk of failing through Artificial Intelligence

A worrying trend that recently affected the university system is characterized by the students’ dropout. Universities usually link the problem with some aspects as study program, structure, and organization of the examinations, that require more involvement from students, that negatively affect their motivation. Even if universities make some improvement actions, as tutoring, to provide students the best approach for their studies aimed at promoting academic success and avoiding university drop-out, sometimes they don’t seem to achieve the results expected. It can happen that the factors which led students to drop-out cannot be related to their approach in the study but can be due to the students’ engagement and social interaction. Universities find out these factors only after students’ drop-out, checking their activities and attendance only at the end of the academic year, too late for avoiding severe consequences. This work reports a possible solution to this problem by exploiting Artificial Intelligence methods based on machine learning, firstly applying clustering to group students according to their behavior and then implementing a classification model to predict students at risk. Once checked the accuracy of the machine learning models, the application designed and realized in this work has been plugged in an online platform to allow the universities’ staff to easily execute the software supporting the students to achieve their goals in terms of engagement and learning outcomes. This is a contribution to reduce university drop-out, with possibility to improve the proposed application by user feedbacks and large amount of data collected.