Optimized Naïve Bayesian Algorithm for Efficient Performance

Abstract

Naïve Bayesian algorithm is a data mining algorithm that depicts relationship between data objects using probabilistic method. Classification using Bayesian algorithm is usually done by finding the class that has the highest probability value. Data mining is a popular research area that consists of algorithm development and pattern extraction from database using different algorithms. Classification is one of the major tasks of data mining which aimed at building a model (classifier) that can be used to predict unknown class labels. There are so many algorithms for classification such as decision tree classifier, neural network, rule induction and naïve Bayesian. This paper is focused on naïve Bayesian algorithm which is a classical algorithm for classifying categorical data. It easily converged at local optima. Particle Swarm Optimization (PSO) algorithm has gained recognition in many fields of human endeavours and has been applied to enhance efficiency and accuracy in different problem domain. This paper proposed an optimized naïve Bayesian classifier using particle swarm optimization to overcome the problem of premature convergence and to improve the efficiency of the naïve Bayesian algorithm. The classification result from the optimized naïve Bayesian when compared with the traditional algorithm showed a better performance

Keywords: Data Mining, Classification, Particle Swarm Optimization, Naïve Bayesian