Bibliographic citations
Daza, A., (2019). Un modelo híbrido basado en redes neuronales y árboles de decisión para predecir la deserción estudiantil en la educación superior privada [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/19880
Daza, A., Un modelo híbrido basado en redes neuronales y árboles de decisión para predecir la deserción estudiantil en la educación superior privada [Tesis]. PE: Universidad Nacional de Ingeniería; 2019. http://hdl.handle.net/20.500.14076/19880
@phdthesis{sunedu/3503031,
title = "Un modelo híbrido basado en redes neuronales y árboles de decisión para predecir la deserción estudiantil en la educación superior privada",
author = "Daza Vergaray, Alfredo",
publisher = "Universidad Nacional de Ingeniería",
year = "2019"
}
One of the main problems facing Private Universities at national and global level in 2018 is the Private University desertion, which has been partially investigated and according to the research carried out, is in the first semester by 40% which given the seriousness of the study, many experiments have been carried out independently, making use of data mining techniques: neural networks, decision trees, SVM, cluster. As well as statistical methods: logistic regression, discriminant analysis, structural equations, where good results have been obtained but with little precision. This research will help the educational institutions, to identify students who are going to drop out of the University, in the early, middle and late ages with high precision and thus take the appropriate measures to reduce the high dropout rate. In the research, the commercial tool rapid miner Studio was used, where the training of decision tree models, neural networks and the proposed hybrid model was carried out with 1761 student data and 26 factors, then the test was performed with 100 new data that were not used for training, where an accuracy of 87% was obtained with decision trees, with neural networks 91% accuracy was obtained and finally a proposed hybrid model was used (decision trees and networks neuronal) where an accuracy of 98% was obtained, obtaining better results than the aforementioned techniques independently, after having made the comparisons in relation to the accuracy. With the obtained result will be able to identify the students who desert with high precision.
This item is licensed under a Creative Commons License