Clasificadores por redes Bayesianas (Spanish text).

摘要

A Bayesian network is a compact representation of joint probability function. Formally, a Bayesian network is an acyclic directed graph in which each node represents a random variable and the relationships of dependence and independence are established by the structure of the network. In this thesis the performance of some classifiers constructed in basis of estimation of Bayesian networks is evaluated. The classifiers considered are Naive Bayes, TAN and Bayesian multinets. The evaluation is carried out on 20 datasets for the UCI Machine Learning Repository. Our results show that classifiers based on the structure of a Bayesian network have a good predictive performance.