A new Exponential Naive Bayes Classifier with Fuzzy Parameters
Anny K. G. Rodrigues, Thiago Vinicius Vieira Batista, Ronei Marcos de Moraes, Liliane Santos Machado · 2016
It is difficult to classify data which follow the exponential distribution. For this reason, in this paper we propose a new classifier based on that distribution, named Exponential Naive Bayes network with Fuzzy Parameters (ENB-FP), where those parameters are given by fuzzy numbers. In order to know performance of ENB-FP, tests using data from six different statistical distributions were performed. ENB-FP have achieved an agreement degree of “almost perfect”, according to Kappa Coeficient, in five of them. A brief comparison with another fuzzy classifier recently proposed was performed as well. According to the Kappa Coefficient, the ENB-FP outperformed that classifier when using data from exponential distribution and provided a competitive approach for the other five distributions.