气候平均值和方差的次序统计估计量order statistics ofm ean and .pdf

30 5 Vo.l 30 N o. 5 2007 10 Journal ofN anjing Institute ofM eteorology Oct. 2007 熊安元, 王颖 (, 100081) : 给出了估计样本平均值和方差的次序统计量 中值伪标准差加权均值加权标准差的数 学描述, 对其统计估计性 一致性无偏性和有效性进行了模拟评估, 评估分析结果表明: 这些 次序统计量具有 一致性和无偏性, 有效性在正态总体下低于传统统计量, 在偏态总体下接近或超过 传统统计量, 加权均值和加权标准差的有效性高于中值和伪标准差, 接近传统统计量模拟和实例 均表明, 与传统的均值和标准差估计量相比, 当资料存在特异值时, 次序统计量具有明显的优越性 : 次序统计量; 平均值; 方差 : P468. 0 : A : 1000-2022( 2007) 05-06 4-05 Order Statistics ofM ean and Variance for Climatic Data X IONG A n-yuan, WANG Y ing ( NationalM eteorological Inform ation Center, Beijing 100081, China) Abstract: In th is paper, themathem atical description of these techn iques involves all estmi ation of param- eters such asmed ian, pseudo-standard deviation, w eightedmean, w eighted standard deviation. Their statis- tical estmi ation performances such as consistency, unb iasedness and valid ity are assessed. The results show that these order statistics are consistent and unbiased. The valid ity is under the traditional statistics for normal d istribution, nears or exceeds for skew ed d istribution. The validity ofw eightedmean andw eigh- ted standard deviation exceedsmed ian and pseudo-standard deviation, and is close to traditional statistics. The smi ulation and examp le both prove that order statistics are suprior as compared w ith the traditional statisticswhen outliers are presented in data. K ey words: order statistics; mean; variance 0 , , , , , ( re- [ 1] , , sistant o