Description

We present NB and Feature Selection are two very essential algorithms for achieving this task because of their ease and high performance. In this paper, we present a hybrid classification algorithm founded on NB and Feature Selection, simply NB-Feature Selection. In NB-Feature Selection, the class chance estimates of NB and Feature Selection are weighted according to their classification accuracy on the training data. The experimental results show that NB-Feature Selection significantly outperforms NB and Feature Selection in terms of classification accuracy. A Bayesian network consists of a structural model and a set of conditional probabilities. The structural model is a directed acyclic graph in which nodes represent attributes and arcs represent attribute dependencies. Attribute dependencies are quantified by conditional probabilities for each node given its parents. Classification is one of the fundamental problems in data mining and knowledge discovery. In classification, the goal is to learn a classifier from a given set of instances with class labels, which correctly assigns a class label to a test instance. The performance of a classifier is usually measured by its classification accuracy (the percentage of instances correctly classified). Classification has been extensively studied and various learning algorithms have be developed, such as decision tree and Bayesian network, that can be categorized into two major approaches: probability-based approach and decision boundary-based approach. In this paper, we focus on the probability-based approach. Decision tree is one of the most widely used classification models. It classifies a test instance by sorting it down the tree from the root node to one leaf node, which provides the classification of this instance via simply voting. Each node in the tree specifies a test of one attribute of the instance