论文部分内容阅读
应用Bootstrap法,对现有样本进行再抽样,得到Bootstrap样本,并计算Bootstrap样本特征值及主分量得分系数,从而可得总体特征值、主分量得分系数的Bootstrap估计。可以将主分量得分系数的Bootstrap估计作为统一权衡标准,对任何样本指标进行加权压缩,得到若干个主分量,然后根据所得主分量利用聚类分析的方法对个体分类,达到初保达标分类评价的目的。本文应用的方法可以使现有样本所提供的信息得到充分利用,从而导出适用于对所有个体评判的一般规则,且不受总体分布类型的影响,因此该方法是比较稳健优越的方法,尤其适合于大量抽样资料分析。
Using Bootstrap method, the existing samples are re-sampled to obtain Bootstrap samples, and Bootstrap sample eigenvalues and principal component score coefficients are calculated, so as to obtain the Bootstrap estimation of the overall eigenvalues and the principal component score coefficients. The Bootstrap estimate of the score of the principal component can be used as a uniform trade-off criterion, and any sample index is weighted and compressed to obtain several principal components. Then the individual principal component is used to classify the individual using the cluster analysis method to achieve the evaluation of the primary assurance standard. purpose. The method used in this paper can make full use of the information provided by the existing samples, thereby deriving the general rules applicable to the evaluation of all individuals, and is not affected by the overall distribution type, so this method is a more robust and superior method, especially suitable Analysis of a large number of sampling data.