Authors:
Jorge Rodas-Silva
1
and
Jorge Parraga-Alava
2
Affiliations:
1
Facultad de Ciencias e Ingeniería, Universidad Estatal de Milagro, Cdla. Universitaria Km 1 1/2 vía Km 26, Milagro, Ecuador
;
2
Facultad de Ciencias Informáticas, Universidad Técnica de Manabí, Avenida José María Urbina, Portoviejo, Ecuador
Keyword(s):
Educational Data Mining, Predicting Academic Performance, Online Education, Low-Income Students.
Abstract:
The success of higher education institutions in the online learning environment can be measured by the performance of students. Identifying backgrounds or factors that increase the academic success rate of online
students is especially helpful for educational decision-makers to adequately plan actions to promote successful outcomes in this digital landscape. In this paper, we identify the factors that contribute to the academic
success of students in public Ecuadorian online universities and develop a predictive model to aid in improving
their performance. Our approach involved five stages: data collection and description, which involved gathering data from universities, including social demographic, and academic features. In preprocessing, cleaning,
and transforming the data to prepare it for analysis was performed. Modeling involved applying machine
learning algorithms to identify patterns and key factors to predict student outcomes. It was validated in the
next stage where t
he performance of feature selection and predictive model was tackled. In the last stage, were
interpreted the results of the analysis about the factors that contribute to the academic success of low-income
students in online universities in Ecuador. The results suggest that the grade in the leveling course, the family
income, and the age of the student mainly influence their academic performance. The best performances were
achieved with Boruta + Random Forest and LVQ + SVM, reaching an accuracy of 75.24% and 68.63% for
binary (Pass/Fail) and multiclass (Average/Good/Excellent) academic performance prediction, respectively.
(More)