Estudio del estado del arte sobre la predicción de deserción universitaria usando machine learning

Abstract

The main consequence of university desertion is the increase in the number of students with incomplete higher education. Therefore, predictive techniques have been implemented that have helped to detect patterns that are useful for the study of the subject. There are several techniques used for predictive analysis such as: Logistic Regression, Bayesian Analysis, Decision Trees, among others, which have become extremely important when it comes to yielding favorable results that help to keep statistical control over student cases. deserters. It is important to point out that the use of predictive techniques such as data mining where it can be visualized the reasons why students drop out according to patterns expressed statistically so that institutional authorities can make decisions based on this information and reduce the dropout rate of university students. This article collects information on the different prediction algorithms used in the subject of university dropout and analyzes the data on techniques, algorithms, datasets, used for the prediction of university dropout, with the aim of finding the technique with the best performance at the moment. to carry out the study on university desertion through evaluation metrics such as Accuracy, Recall and F-Score.