An objective general index for multivariate ordered data
An objective general index for multivariate ordered data
复制标题
DOI：
10.1016/j.jmva.2016.02.005
复制
发表时间：
2016-05-01
影响因子：
1.6
通讯作者：
Sei, Tomonari
中科院分区：
文献类型：
--
作者：
Sei, Tomonari
当变量具有一定的顺序时，多变量的定量数据通常被归纳为一个总指数，作为加权和。虽然标准化得分之和是一个合理的指标选择，但它可能与某些变量呈负相关。本文构造了一个与所有变量均呈正相关的综合指数。运用该指标对十项全能评分的公平性进行了研究。有序分类数据的量化也进行了讨论。定量限表征高斯分布。(C)2016 Elsevier Inc. All rights reserved.
Multivariate quantitative data are often summarized into a general index as a weighted sum when the variates have a prescribed order. Although the sum of standardized scores is a sensible choice of index, it may have negative correlation with some of the variates. In this paper, a general index that has positive correlation with all the variates is constructed. The index is applied to study the fairness of decathlon scoring. Quantification of ordered categorical data is also discussed. The limit of quantification characterizes the Gaussian distribution. (C) 2016 Elsevier Inc. All rights reserved.