Predicción del rendimiento académico por medio de técnicas de inteligencia artificial de los estudiantes de bachillerato en las Unidades Educativas fiscales del distrito educativo 05D06 (Salcedo - Ecuador)

Abstract

The aim of this research was to determine the mathematic models that predict the academic performance of high school students in public educational institutions in Salcedo. The variables of the study were determined by books and academic papers which indicate the factors that affect, predict, or explain academic performance; obtaining 21 variables. We had access to the information of 7 public educational institutions in the countryside with a total of 279 students, and 3 public educational institutions in the city with a total of 402 students through surveys and their grades. Data were organized and elaborated by Excel. The information was analyzed by the software R; having used the multiple linear regression with the adjusting R square, and the standard error to determine which models are good indicators of academic performance. On the other hand, for the logistic regression, the matrix of confusion with the percentage of correct answers was used to determine if the model was a good indicator. In the multiple linear regression of the institutions in the countryside, it was determined 6 predicting variables with an adjusting R square of 0.8057, and a standard error of 0.2373. In the institutions of the city, it was obtained 7 variables with a standard error equal of 0.3, and an adjusting R square of 0.7525. In the logistic ordinal regression of the institutions in the countryside, it was determined a model with 5 predicting variables and a percentage of correct answers of 63.09524%. For the institutions of the city, there are 6 predicting variables and a percentage of 52.45902. We conclude that the logistic regression was not a good tool to predict the academic performance of high school students in public educational institutions in Salcedo. The multiple linear regression is a good tool for predicting academic performance, it is demonstrated through other research, and it is confirmed by the mentioned results of this investigation.