ANALYSIS OF BAYESIAN CLASSIFIER ACCURACY

Felipe Schneider Costa,Silvia Modesto Nassar,Maria Marlene De Souza Pires

doi:10.3844/jcssp.2013.1487.1495

Felipe Schneider Costa, Silvia Modesto Nassar + Show 1 more

Open Access

https://doi.org/10.3844/jcssp.2013.1487.1495

Copy DOI

Abstract

The naive Bayes classifier is considered one of the most effective classification algorithms today, competing with more modern and sophisticated classifiers. Despite being based on unrealistic (naive) assumption that all variables are independent, given the output class, the classifier provides proper results. However, depending on the scenario utilized (network structure, number of samples or training cases, number of variables), the network may not provide appropriate results. This study uses a process variable selection, using the chi-squared test to verify the existence of dependence between variables in the data model in order to identify the reasons which prevent a Bayesian network to provide good performance. A detailed analysis of the data is also proposed, unlike other existing work, as well as adjustments in case of limit values between two adjacent classes. Furthermore, variable weights are used in the calculation of a posteriori probabilities, calculated with mutual information function. Tests were applied in both a naive Bayesian network and a hierarchical Bayesian network. After testing, a significant reduction in error rate has been observed. The naive Bayesian network presented a drop in error rates from twenty five percent to five percent, considering the initial results of the classification process. In the hierarchical network, there was not only a drop in fifteen percent error rate, but also the final result came to zero.

Highlights

Many tasks, including fault diagnosis, pattern recognition and forecasting can be seen as classification (Cheng and Greiner, 1999)
The naïve Bayesian network presented a drop in error rates from twenty five percent to five percent, considering the initial results of the classification process
The matrix columns receive response values generated by the network and lines receive the output class values according to gold standard (Marsland, 2009)

Summary

Introduction

Many tasks, including fault diagnosis, pattern recognition and forecasting can be seen as classification (Cheng and Greiner, 1999). The classification is a base task in data analysis and pattern recognition which requires the construction of a classifier, that is, a function that assigns a class tag to examples described by a set of variables. The inference of classifiers on data sets with pre-classified cases is a central problem in machine learning. Several approaches to this problem are based on functional representations such as decision trees, neural networks and rules (Friedman et al, 1997). In the study by (Lee et al, 2011) weights are assigned to the variables of the data set by using the Kullback-Leibler measure

Objectives

Methods

Results

Conclusion