逆高斯分布变异系数和尺度参数的统计推断研究

构造了逆高斯分布中变异系数的广义枢轴量,给出了一种参数的区间估计方法,并与MOVOER(method of variance of estimates recovery)和Bootstrap方法进行比较;给出了多总体下尺度参数两两差的同时置信区间.模拟结果表明:在中、小样本情况下,所给的广义置信区间其覆盖概率接近置信水平,平均区间长度较短,优于MOVOER方法与Bootstrap方法;对于多总体下尺度参数两两差的同时置信区间,所给出的三种同时置信区间,其覆盖率在置信水平附近,具有良好的频率性质.

Statistical Inference of Coefficient of Variation and Scale Parameters in Inverse Gauss Distribution

In this paper,a generalized pivotal variable of the coefficience of variation in inverse Gauss distribution is constructed,an interval estimation method of parameters is given,comparing it with MOVOER(method of variance of estimates recovery)and Bootstrap methods;a concurrent confidence interval for the difference between two subscale parameters of Multi-population is given.The simulation results show that the coverage probability of the generalized confidence interval given in this paper is close to the confidence level,and the average confidence length is shorter,which is better than MOVOER method and Bootstrap method.For the Simultaneous Confidence Intervals of Multi-population subscale parameters,the three concurrent confidence intervals given in this paper have good frequency properties with coverage near the confidence level.