论文部分内容阅读
对全国农残水平测试中毒死蜱、氯氰菊酯数据进行统计分析,在数据统计分布特征研究基础上,使用内核密度估计进行数据多态性分析,使用bootstrap模拟取样法对数据样本值重复取样,以获得稳健的水平测试样品待测物含量代表值估计、标准误差及置信区间描述,证明以bootstrap模拟取样法获取的均值与标准偏差作为有限单次样本代表值是合理、有效的,解决了四分位稳健统计方法对非正态多态分布代表值估计不稳定问题及取样理论中取样样本数限制的瓶颈,为能力验证计划指定值的获取提供了一种新方法。
Based on the statistical distribution of chlorpyrifos and cypermethrin in the national pesticide residue level test, the data were analyzed by kernel density estimation, and the sample data were repeatedly sampled using the bootstrap simulation sampling method to obtain a stable The standard error and the confidence interval description of the test samples were used to prove that the mean and standard deviation obtained by the bootstrap simulation sampling method are reasonable and effective as the representative value of the finite single sample, The instability of statistical methods for instability estimation of representative values of non-normal polymorphic distributions and the bottleneck of the limitation of the number of samples in sampling theory provide a new method for obtaining the specified value of proficiency testing plan.