期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2014
卷号：5
期号：2
页码：1398-1402
出版社：TechScience Publications
摘要：Discretization plays a significant role during the transformation of continuous attributes into discrete values in the pre-processing step in data mining, significant attention in the machine learning community. Recently, many researchers have provided numerous discretization methods applied to classification problems. Moreover, these are potential to reduce the dimensionality of the data during the classification accuracy, due to this reason which may bias in classifier accuracies and results. In this paper, we propose the new discretization algorithm based on a popular statistical technique called a z-score and particle swarm optimization technique for feature selection to identify the highly influenced features using wrapper-based feature subset selection. Machine learning based algorithm (C4.5 decision tree) to generate highly accurate decision rules. Empirically, we conduct an experimental study on benchmark continuous data sets with different type of machine learning classifiers. The result shows high performance computed in terms of accuracy and inconsistency has been tested with a Wilcoxon signed-rank test.
