USING ARTIFICIAL INTELLIGENCE TECHNIQUES FOR ESTIMATING THE PERFORMANCE OF GRADUATE STUDENTS AT THE DEPARTMENT OF MATHEMATIC SCIENCES

Ircílio Chissolucombe, Rodrigo Souza, Hugo Oliveira, Romério Lima

Abstract


This paper presents a study in the School ofTechnology of an Institution of Higher Education of theFederal District - Brasília, to estimate the performance ofstudents in the disciplines of Mathematic Sciences. Adatabase of 700 pairs of examples and techniques ofartificial intelligence, fuzzy k-means and Neural Networkswere used. The training data and test were separated usingthe technique of fuzzy k-means clustering, which allowsgrouping elements with similar characteristics. Separated20% data to compose the test set and the remaining 80%were part of the training set. It was used for training theneural network algorithms and resilient backpropagationwith Bayesian regularization. The best results were obtainedwith the resilient backpropagation algorithm. Thecomputational tool that was used is MATLAB (Mathworks,2006). In the results it was obtained a correlation coefficientof R2 = 0.98 and that will aid in pedagogical decisions.

Full Text: PDF (Português (Brasil))

Refbacks

  • There are currently no refbacks.


Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.

----------------------------------------------------------------------

Indexing

Logotipo do IBICT

----------------------------------------------------------------------

Scientific Societies and Directories

Logotipo COPEC Logotipo SHERO Logotipo da Capes

----------------------------------------------------------------------

Follow Us

Logotipo facebook Logotipo LinkedIn Logotipo Twitter