点击次数：

论文类型：期刊论文

发表时间：2013-03-01

发表刊物：Journal of Information and Computational Science

收录刊物：EI、Scopus

卷号：10

期号：4

页面范围：949-957

ISSN号：15487741

摘要：Data stream classification has drawn increasing attention from the data mining community in recent years. In many real-world applications, concept drift usually seriously affects the performance of classification. In order to handle it, we propose a novel data stream classification framework, which extracts the concept from each data block, finds the best suitable classifier from the classifier pool for classification. In the experiment, two kinds of data sets, synthetic and real-world, are employed to evaluate the validity of the proposed model. The experimental result shows the proposed model can improve the accuracy of data stream classification under concept drift. ? 2013 by Binary Information Press.