(2185) ARTIFICIAL NEURAL NETWORKS FOR THE PREDICTION OF STUDENTS ACADEMIC PERFORMANCE

摘要

The implementation of mathematical methods for the solution of social problems has been doing formany years. Just like there still few studies where the data mining techniques are applied or artificialneural networks for obtaining, analysis and modelling of data, the tendency of the used of these toolsin combination of computer science, has been increased through software such as SAS, SPSS,Matlab among others that were manufactured particularly by researchers to solve specific problems.This paper describes the methodology to apply techniques of artificial neural networks (ANN) for theprediction of academic performance. This is intended to make a classification of the aspirants to enterto a university career in different levels according to the probability of reaching a given performancecategory (e. i. excellent, good, acceptable, insufficient, and poor).Increase discharge indicators, titling and terminal efficiency;;decrease the dropout rate, especially thefirst and second year of the career, are areas of opportunity in any educational institution, regardlessof its prestige, nationality and area of knowledge. This paper proposes a methodology based on arational technique to improve these indicators.Implement artificial neural networks for the prediction of academic performance is a field that still littleexplored, there are some isolated pieces of work with this application. In this work is carried out anexhaustive review of the literature, It is a methodological proposal and concludes with a comparisonwith other methods of prediction as multiple linear regression and data mining.This research aims to expand in future work to predict shortcomings of the students in various aspectsof the courses during the academic career and suggest actions to avoid them.