A review on the prediction of students’ academic performance using ensemble methods

Danna Lorena Morales Rodríguez

Universidad Distrital Francisco José de Caldas

Joan Alejandro Caro Silva

Universidad Distrital Francisco José de Caldas

Leonardo Emiro Contreras Bravo

Universidad Distrital Francisco José de Caldas

Introduction: This article is a product of the research “Ensemble methods to estimate the academic perfor-mance of higher education students”, developed at the Universidad Distrital Francisco José de Caldas in the year 2021, focusing on the review of research work developed in the last five years related to the prediction of academic performance using ensemble algorithms.
Objective: The literature review aims to identify the most used algorithms and the most relevant variables in the prediction of academic performance.
Methodology: A systematic review of the literature was carried out in different academic databases (Science Direct, Scopus, SAGE Journals, EBSCO, ResearchGate, Google Scholar), using search equations built with keywords.
Results: 54 related articles were found that meet the inclusion criteria of the review. Additionally, benefits were found in the application of ensemble methods in the prediction of academic performance.
Conclusion: It was found that the most influential variables in academic performance correspond to the aca-demic factor. The algorithm used that presents the best results is Random Forest; in addition to being the most used. The use of these algorithms is an accurate tool to predict academic performance at any stage of university life, and at the same time provide information to generate strategies to improve dropout and academic retention indicators.

Keywords: ensemble methods, academic performance, machine learning, educational data mining (EDM)
Published
2022-05-10
Downloads

How to Cite

[1]
D. L. Morales Rodríguez, J. A. Caro Silva, and L. E. Contreras Bravo, “A review on the prediction of students’ academic performance using ensemble methods”, ing. Solidar, vol. 18, no. 2, pp. 1–28, May 2022, doi: 10.16925/2357-6014.2022.02.01.
Metrics
File downloads
681
Twitter
1
Captures
  • Readers: 21
Mentions
  • News Mentions: 1