Application of data mining techniques to predict the academic performance of the students of the ‘Lic. Angélica Villón L.’

Abstract

One of the goals of the ‘Lic. Angélica Villón L.’ school is to improve the academic level of its students. For this, having tools that allow the availability of academic information for decision-making is essential. For this purpose, elements such as the student population, the grades achieved at each level, family support, among others, are prioritized, all cataloged as the student's academic performance. However, the evaluation of this indicator is limited due to the influence of several factors that need to be processed in an integrated way depending on their level of influence. The research carried out is observational, exploratory level; shows the use of business intelligence tools as support for making decisions; a data warehouse is created as a unified repository through ETL processes; Supervised learning models such as vector support machines, neural networks and regression decision trees are trained to predict academic performance. Historical student data is the source for the application of the models. Finally, the model with the best precision is identified through valid metrics in the context of regression analysis.