Regresar

Red bayesiana jerárquica para analizar datos educativos

Abstract:

This document presents a proposal to implement a cluster method that best matches the educational data. The use of probabilistic graphic models in the field of education has been considered for this research. But the problem with these general learning procedures comes from the presence of a high number of variables that measure different aspects of the same concept. In this case, we have that all variables have some degree of dependence between them, without a true causal structure. Therefore, a new procedure is presented that makes a hierarchical grouping of the data while learning a joint probability distribution. We try to make a test for each case where the probability is measured in each model using propagation algorithms. Then, the probability logarithm is applied to each case and the results are added in each model to determine the best fit for the proposed one. The method is applied to the analysis of a data set of educational data: evaluation of students of the Quevedo State Technical University (Ecuador).