多类不平衡数据上的分类器性能比较研究

不同的基分类器对不同分布类型的多类别不平衡数据的适应性存在较大差异。为此，针对分类器的选用问题，在分析比较准确率(ACC)及曲线下面积(AUC)的评价标准基础上，选择基于AUC的分类器评价方法，将支持向量机、决策树和贝叶斯分类器应用于标准数据集中，并采用AUC来评价结果，得出相关结论：在多类不平衡数据上，贝叶斯是最好的基分类器，且SVM分类器存在一定改进空间。

Comparison Research of Classifier Performance on Multi-class Imbalanced Data

On different distribution of multi-class imbalanced data,different base classifiers have different adaptability.To aim at the classifier selection problem,based on the analysis and comparison of the evaluation criteria of Accuracy（ACC） and Area Under the ROC（AUC）,it chooses AUC to evaluate a classifier and draws the conclusion that Bayesian classifier is the best one and SVM classifier has greater room for improvement,through the experiments on dealing with many different distribution unbalanced data from standard database.