• 主题
高级搜索  >
历史搜索 清空历史
首页> 美国政府科技报告 >Modified Goodness-Of-Fit Tests For The Inverse Gaussian Distribution With TwoUnknown Parameter
【24h】

Modified Goodness-Of-Fit Tests For The Inverse Gaussian Distribution With TwoUnknown Parameter

机译：具有两个未知参数的逆高斯分布的修正拟合优度检验

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Modified Kolmogorov- Smirnov (KS), Anderson-Darling (AD), Cramer-von Mises (CV),Kupier (V), and Watson (W) goodness-of-fit tests are generated for the inverse Gaussian distribution with unknown parameters. The inverse Gaussian parameters are estimated by maximum likelihood estimation. A Monte Carlo simulation of 50,000 repetitions is used to generate critical values for sample sizes of 5 through 50 with an increment of five, sample sizes of 60 through 100 with an increment of 10, and 24 different values of the inverse Gaussian shape parameter. A 50,000-repetition Monte Carlo power study is carried out using data with sample sizes of 5 through 100 from five alternate distributions for the five EDF tests for significance levels of 0.01,0.05,0.10, 0.15, and 0.20. For sequential tests, power studies are performed for the significance levels produced by combining two EDF tests. Power studies corresponding two both cases are presented in tables and graphs. The power studies showed that the tests arc excellent in discriminating between the inverse Gaussian and distributions such as the gamma, exponential and uniform that are very different in shape. However, they are relatively unable to discriminate between the inverse Gaussian distribution and distributions that are similar in shape such as the lognormal and certain Weibull distributions with shape similar to the particular inverse Gaussian. The AD test has the highest power in most cases studied. A functional relationship is identified between the modified KS, AD, CV, V, and W test statistics, sample size, and the inverse Gaussian shape parameter. The critical values are found to be a non-linear function of the shape parameters and sample sizes for the significance levels of 0.01, 0.05, 0.10, 0.15, and 0.20.

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号