药品质量控制中常用异常值判别方法辨析
Discrimination of the outlier tests commonly used in pharmaceutical quality control
分类号：
出版年·卷·期（页码）：2013,33 (11):0-0
DOI:
10.16155/j.0254-1793.2017.01.01
-----摘要：-------------------------------------------------------------------------------------------
目的: 正确地运用各种异常值判别方法，科学地进行数据分析。方法: 用药品检定实验者2个常见数据集实例，对常用异常值检验方法的敏感性进行比较；对异常值判断的重要性、判别标准等进行分析。结果和结论: 各种异常值判断方法都有其局限性。实验结果的准确一致是判断异常值分析正确性的前提；在实际运用中，要严格按照分析步骤进行操作，最好采用多种异常值检测方法同时分析；当选用一种方法时，应根据具体情况咨询专业统计人员进行选择。
-----英文摘要：---------------------------------------------------------------------------------------
Objective: To use the outlier tests correctly for scientific data analysis. Methods: The sensitivity of the common outlier tests were compared by using data sets of two common experiments and the importance and judging criteria of the outlier test were analyzed. Results and Conclusion: Each outlier testing method has its own limitations.The accuracy and consistency of the experimental results are the premise to determine whether the outlier analysis is correct or not.In practical use,it is necessary to operate in strict accordance with the analysis steps,preferably to use a variety of outlier test methods for analysis.When selecting an outlier method to be used in certain practice,it is recommended to consult a professional statistician.
-----参考文献：---------------------------------------------------------------------------------------
1 Josef TP,Linsinger J,Kandler W,et al.The influence of different evaluation techniques on the results of interlaboratory comparisons[J].Accred Qual Assur,1998,3:322
2 Ben-Gal I.Outlier detection In:Maimon O,Rockach L.Data Mining and Knowledge Discovery Handbook:A complete Guide for Practitioners and Researchers[M].Norwell,Massachusetts,USA:Kluwer Academic Publishers,2005.1
3 Hrominski KC,Tkacz M.Comparison of outlier detection methods in biomedical data[J].J Med Inf Technol,2010,16:89
4 Graf H,Hampel FR,Talter J-D.The problem of unsuspected serial correlations[J].Lect Notes Stat,1984,26:127
5 Solak MK.Detection of Multiple Outliers in Univariate Data Sets [EB/OL].[2009-06-05] .http://www.lexjansen.com/pharmasug/2009/sp/sp06.pdf
6 WANG Wen-zhou(王文周).t test ─the superlative test to discard abnormal values with σ unknown (未知σ,t检验法剔除异常值最好)[J].J Sichuan Univ Sci Technol(四川工业学院学报),2000,19(3):84
7 ZHENG Yong-xi(郑用熙).Statistical Methods in Analytics(分析化学中的数理统计方法)[M].Beijing(北京):Science Press(科学出版社),1991.104,312
8 TAN De-jiang(谭德讲),MA Shuang-cheng(马双成).Statistical Applications in Pharmaceutical Regulatory Science (药品监督与检定中的统计学应用)[M].Beijing(北京):China Science and Technology Press(中国科学技术出版社),2011
9 Shafer JL.Analysis of Incomplete Multivariate Data[M].London:Chapman & Hall Publishers,1997.262
欢迎阅读《药物分析杂志》！您是该文第 1328位读者！
