利用Origin对数据异常值的剔除方法进行比较  

Origin Used in Comparison the Methods of Eliminating the Excrescent Data

作　　者：张璇[1] 程敏熙[1] 肖凤平[1] 

机构地区：[1]华南师范大学物理与电信工程学院,广州510006

出　　处：《实验科学与技术》2012年第1期74-76,118,共4页Experiment Science and Technology

摘　　要：介绍了采用拉依达准则、肖维勒准则、格拉布斯准则以及狄拉克准则4种剔除异常数据的方法,简述了各准则的数理统计原理、处理步骤,并借助Origin软件对处理某次盐液体电导率实验的数据,对比了各准则剔除异常值后的处理结果.指出在大样本情况下较适合用拉依达准则,而肖维勒准则和狄拉克准则适用于样本容量较小的情况,格拉布斯准则则适用性和剔除结果较为普遍。Four criterions are introduced in this paper to eliminate and distinguish the abnormal values,they are PauT a,Chauv-enet,Grubbs,and Dixon.Their mathematic statistics principles and concordances are also resumed in this paper.Origin software are used to deal with experimental data of conductivity of solution and to compare the processing results.It pointed out that PauT a and Chauv-enet are suitable for large sample and Grubbs,Dixon is suitable for small sample.The Grubbs criterion is in common use.

关 键 词：ORIGIN软件 异常值 拉依达准则 肖维勒准则 格拉布斯准则 狄克逊准则 

分 类 号：TP37[自动化与计算机技术—计算机系统结构] O4-33[自动化与计算机技术—计算机科学与技术]