泊松分布参数的最短置信区间  

The Shortest Confidence Interval of Parameter λ in the Poission Distribution

作　　者：岑忠[1] 丁勇[1] 

机构地区：[1]南京医科大学基础医学院,210029

出　　处：《中国卫生统计》2010年第2期133-135,共3页Chinese Journal of Health Statistics

摘　　要：目的根据抽样结果估计泊松分布参数λ的最短置信区间。方法证明了参数λ最短置信区间的存在和唯一性,利用搜索法用Matlab编程求得最短置信区间,将最短置信区间的长度与用一般的方法求得的置信区间的长度进行了比较。结果对置信度0.90、0.95、0.99和样本总计数1～50给出了最短置信区间表,可作为统计用表使用。结论最短置信区间估计精度比用一般的方法的精度有显著提高,用最短置信区间作为参数λ的区间估计更精确。Objective To estimate the shortest confidence interval （CI） of parameter λ in the Poission distribution by Sampling result.Methods Bases on the unique existence of shortest CI of λ,searching approach was accepted in Matlab program to find the shortest CI.A comparative analysis was used between the lengths of shortest CI and CI in a general sense.Results A table of shortest CI of λ was given for degree of confidence 0.90,0.95,0.99 and sampling total count 1-50,it could be used as a statistic table.Conclusion Precision of the CI estimation of parameter λ was enhanced compared to the CI in a general sense.It is meaningful to use the shortest CI of λ as its interval estimation.

关 键 词：泊松分布 最短置信区间 搜索法 

分 类 号：R195[医药卫生—卫生统计学]