ROC分析技术在机器学习中的应用

Application of ROC analysis in machine learning

ROC(受试者工作特征)分析技术是一种用来衡量分类算法和图示它们性能的技术。与传统的正确率相比,ROC分析更能够全面地描述分类算法的分类性能。该方法具有可信度高,描述客观精确,特别是不受数据环境影响等优势。对国内外这一方法的研究成果进行了较为系统地介绍,详细分析了它的优缺点,最后对这一技术的发展进行了展望。

Receiver Operating Characteristics（ROC） analysis is a technique for organizing classifiers and visualizing their performance.Comparing with general accuracy,ROC could describe the classify capability adequately and be adopted in all conditions. This method has many characteristic such as high reliability,object and accurate describe and especially it cannot be influenced by the data environment.In this paper,we systematically discuss the research achievement and introduce the advantages and shortcomings about ROC analysis.In the end we look forward to its development.