Examinando por Autor "Espinosa Pinos, Carlos Alberto"
Mostrando 1 - 1 de 1
Resultados por página
Opciones de ordenación
- ÍtemModelo predictivo para el rendimiento académico en la asignatura de matemáticas(Fundación Universitaria Los Libertadores. Sede Bogotá., ) Espinosa Pinos, Carlos Alberto; Gonzalez Veloza, John FredyThe purpose of this article was to apply Machine Learning algorithms to classify the achievement in mathematics of secondary school students in Ecuador, which allowed determining the attributes of the studied database that best contribute when proposing a predictive model. Three modles were developed to identify the presence of behavior patterns such as passing or non-passing achievement, analyzing numerical variables such as grades in exams for other subjects or for admission to higher eduaction, and categories such as financing of the institution, student ethnicity, sex between other. The applied methodology refers to 7 of the 8 steps used in data science proposed by SUNK With the support of the Python library sklearn, the generation of the models was proposed. As a result of the work, the best model corresponding to a random forest was selected with 92% in precision, accuracy in addition to having 98% in memory or Recovery and an Accuracy of 97%. They identified attributes to the model mentioned as: higher education entrance exam grade, undergraduate exam and achievement grades in linguistic, scientific and social studies domain. Additionally, it was possible to improve the balance in the database by making a cut with the score of 8 and consequently a better interpretation of the results