【作者】 杨波； 秦锋； 程泽凯；

【Author】 Yang Bo,Qin Feng,Cheng Zekai School of Computer Science Anhui University of Technology,Ma’anshan,243002,Department of Computer Science Tonghua Teachers College,Tonghua,134002

【机构】 安徽工业大学计算机学院； 通化师范学院计算机系；

【摘要】 预测的准确率已被广泛用于作为分类学习系统预测性能的主要评估标准。然而,它存在着如忽视预测产生的概率信息等诸多缺点和不足之处。文章中、提出并推荐使用一种新的分类学习系统评估度量——AUC(the area under the Receiver Opernting Characteristic curve).它弥补了准确率评估标准的不足,利用了分类器所产生的概率或分类,具有诱人的性质和广阔的应用前景。通过分析、比较得出结论,AUC是优于准确率并应替代准确率的分类学习系统评估度量。更多还原

【Abstract】 Predictive accuracy has been used widely as a main evaluation criterion for predictive performance of classification,learning system.However,it has many shortcomings and disadvantages.For example,it ignores probability estimations that classifiers produce.In this paper,AUC(the area under the Receiver Operating Characteristic curve) as a new measure of classification learning system is recommended and used,Which makes up for the deficiencies of accuracy and makes use of probability estimations or scores that classifiers produce.Its properties are attractive and it will be applied extensively.The analysis shows that AUC is a better measure than accuracy and should replace it in classification learning system.更多还原