摘要
无量纲化的优点是能缩小指标间的量纲差异,缺点是容易造成原始信息损失,科学评价无量纲化方法的利弊有助于寻找不同分布中的最佳无量纲化方法,提高数据的处理质量。文章分别构建无量纲化效果和信息损失情况的评价指标,通过数值模拟实验来比较不同分布中各无量纲化方法的效果和信息损失速率。结果发现,无量纲化过程中利弊并存,不同分布数据适用的无量纲化方法不同,需要综合考虑数据的分布特点和研究的需要来选择无量纲化方法,其中线性无量纲化不改变数据的分布特征,伸缩法不改变数据的变异性,平移法不改变数据的差异性大小,归一化和规范化的无量纲化效果较好,但信息损失速率较快。
The advantage of dimensionless methods is that it can reduce the dimensional difference between indicators,while the disadvantage is that it is easy to cause loss of original information.Scientific evaluation of the advantages and disadvantages of dimensionless methods is helpful in finding the best dimensionless method in different distributions and improving the quality of data processing.This paper constructs the evaluation indexes of dimensionless effect and information loss respectively,and compares the effect and information loss rate of each dimensionless methods in different distributions by a large number of numerical simulation experiments.The results are shown as follows:There are both advantages and disadvantages in the dimensionless process,and the dimensionless methods suitable for different distribution data are different.The dimensionless methods should be selected by comprehensively considering the distribution characteristics of data and research needs.Among them,the linear dimensionless methods do not change the distribution characteristics of data;the scaling methods do not change variation characteristics of data;the translation methods do not change the difference of the data,and the dimensionless effect of normalization is better,but the information loss rate is faster.
作者
李兴奇
高晓红
Li Xingqi;Gao Xiaohong(School of Management and Economics,Chuxiong Normal University,Chuxiong Yunnan 675000,China;School of Mathematics and Computer Science,Chuxiong Normal University,Chuxiong Yunnan 675000,China)
出处
《统计与决策》
CSSCI
北大核心
2022年第10期31-36,共6页
Statistics & Decision
基金
国家自然科学基金资助项目(11261001)
云南省应用基础研究计划青年项目(2017FD152)
楚雄师范学院校级一般项目(XJYB2007)。
关键词
不同分布
无量纲化方法
无量纲化的效果
信息损失速率
弹性
different distributions
dimensionless methods
dimensionless effect
information loss rate
elasticity
