Using Machine Learning for Prediction Students Failure in Morocco: an Application of the CRISP-DM Methodology

Abstract

Student failure prediction is one of the main topics in university learning contexts, as it helps to avoid failure in higher education institutions and provides a basis to make the teaching and learning process more effective, efficient and reliable. The overall aim of this study is to identify students who are susceptible to fail a given university course. This research paper reports the implementation of an Educational Data Mining project based on the CRISP-DM methodology. The data was collected from the APOGEE system of Ibn Tofail University, a form and specifications of the tested courses. The business goal of this paper is to develop a model that can identify students who are susceptible to failure in a given academic course. Such a model helps prevent failure in higher education institutions and provides a basis for making the teaching and learning process more effective, efficient and reliable. Most common machine learning algorithms in the field of Educational Data Mining were used. The results of our research showed that the proposed method was able to achieve an overall accuracy of 97% in predicting students at potential failure.