Prediction of academic performance using the first academic activities of university students and machine learning techniques.

Abstract

The main objective of this study is to design and evaluate academic performance prediction models using machine learning techniques and the first academic activities of university students. The first grades of academic activities of 260 university students were collected to train machine learning techniques to predict the academic performance of 112 students and compare with the real results achieved after the course. An accuracy of predictions of almost 76% was obtained with only a few academic activities at the beginning of the course, which are frequently recorded by teachers, making it easier for this type of study to be replicated in different courses.