Abstract: To improve the classification performance of minority class, this paper combines the advantages of boosting and over-sampling, and presents an over-sampling algorithm based on MCMO-Boost of Adaboost. MCMO-Boost is compared with C4.5, Adaboost and SMOTE, and the results show that MCMO-Boost performs better than others for the classification performance of minority class and the whole data set.

Key words: Unbalanced data set, Over-sampling, Boosting algorithm

摘要： 为了提高不均衡数据集中少数类的分类性能，该文融合了提升和过抽样的优点，提出了基于提升算法Adaboost的过抽样算法MCMO-Boost，并且将其与决策树算法C4.5、提升算法Adaboost和过抽样算法SMOTE进行了实验比较与分析。结果表明，MCMO-Boost算法在少数类和数据集的总体分类性能方面都优于其它算法。

关键词: 不均衡数据集, 过抽样, 提升算法

CLC Number: 

• TP301.6