Open AccessRedes neuronales para predecir el abandono académico en ingeniería
Author(s) -
Humberto José Centurión-Cardeña,
Danice Deyanira Cano-Barrón,
Jesús Sandoval-Gío,
Alfredo Zapata-Gonzále
Publication year - 2020
Publication title -
revista de tecnología y educación
Language(s) - English
Resource type - Journals
ISSN - 2523-0360
DOI - 10.35429/jtae.2020.11.4.1.7
Subject(s) - artificial neural network , context (archaeology) , backpropagation , computer science , index (typography) , artificial intelligence , socioeconomic status , process (computing) , mathematics education , psychology , sociology , world wide web , demography , geography , population , archaeology , operating system
This project describes the design process of an artificial neural network model to predict the risk of dropping out of engineering students throughout their socioeconomic, academic, and personal data sing CRISP-DM methodology. The neural network used in the project considers backpropagation functionality with one hidden layer on data from a context questionnaire and academic data from the students in its CENEVAL’s entrance exam and their academic status after one year in the institution. The data used to train the neural network it’s from 781 records of the last four generations of freshmen year students at the Technological Institute of Motul organized in 48 attributes out of the almost 120 included in the original instrument. The result is a predictive model with a significance level of 75.42% and an F index of 0.6027. This model will be included in the comprehensive tutoring system that is been develop within the organization to monitor the student’s academic performance.