摘要： 针对在维护数据挖掘模型过程中须反复计算数据集、效率较低的问题，基于Ensembles学习思想，研究增量数据集的弱分类器生成方法，根据增量数据集分类器之间的相异度提出新的组合分类算法，分析组合分类器的出错率。实验结果表明，该分类方法是有效的。

关键词: 增量, 分类, Ensemble学习, 组合

Abstract: To maintain a data mining model, it is necessary to repeat related computation for frequent changing data sets, and this will lead to the low efficiency problem. This paper uses Ensemble learning to study generation of weak classifiers on incremental data sets, and presents a new combination algorithm according to dissimilarity of classifiers of incremental data sets. It analyzes the error rate of the combined classifier. Experimental results show the effectiveness of the classification method.

Key words: incremental, classification, Ensemble learning, combination

中图分类号: 

• TP301