Prediction of the opening of university schedule groups using data mining techniques

Authors

Aradiel, Hilario

Acosta, Raul

Mas, Guillermo

Abstract

Universities play an important role in providing educational services, especially in preparing the academic schedule for the semester. This programming is done on a trial and error basis, that is, courses are scheduled and they are not opened because there are few or no students enrolled, seriously affecting educational services. Data mining is the best solution to find hidden patterns and offer suggestions to improve the academic programming of the semester. This article presents a model based on supervised predictive algorithms to accurately predict the number of academic programming groups. In this model, the decision tree classification model was used and the academic records of the students were used. The student's academic record consists of 8 fields: student ID, semester of study, course ID, time group, number of course credits, course approval status and place of study. A total of 44,737 records were collected. The sklearn Python library was used to build this model. Finally, the results show in the hourly group planning indicators reducing from one month to 5 minutes and the percentage of groups indicator that shows us the scheduled courses for each cycle.