THE USE OF MACHINE LEARNING ALGORITHMS TO PREDICT THE ACADEMIC PERFORMANCE OF UNIVERSITY STUDENTS

Abstract

This article examines the possibilities and practical results of using machine learning algorithms to predict the academic performance of students in higher education institutions. Based on an analysis of domestic and foreign research projects, it has been shown that modern data mining methods make it possible to predict students' academic risks with high accuracy – from 85 to 95% – including the likelihood of expulsion, unsatisfactory exam results, and declining academic performance. The paper examines the main approaches to forming sets of features based on students' digital traces in the e-learning environment, academic history, and socio-demographic and psychological characteristics.