指数分布多个异常数据的检验

Detection of Many Outliers from The Exponential Distribution

利用样本分位数构造检验统计量,给出来自于指数分布总体异常数据的一种检测方法.求出了检验统计量精确的概率密度函数和大样本情形下的近似分布,从而得到了检验临界值简洁的近似表达式.检验统计量中的核心统计量——样本分位数,对于异常数据的干扰具有一定的抵抗力,因此该方法可有效地达到检测效果.

A method of detection of outliers from the exponential distribution is given by sample fractile to construct test statistics. The accurate density function and the approximate distribution of test statistics were obtained. In the test statistics, the central statistics-sample fractile has resistance to the influence from the outliers. So, this method can be effectively achieved detection.