两独立样本的非参数检验的一些新进展及展望
Non-Parametric Two-Sample Tests:Recent Developments and Prospects
王柔琳 1潘文亮 2王学钦3
作者信息
- 1. 中国科学技术大学管理学院统计与金融系,合肥,230026
- 2. 中国科学院数学与系统科学研究院,北京,100190
- 3. 中国科学技术大学管理学院国际金融研究院,合肥,230026
- 折叠
摘要
当今数据爆发式地增长,不仅挑战着传统统计推断的理论和方法,而且拓广了统计推断的研究范畴.如何对各种类型的数据进行科学地差异性分析是统计推断研究的一个重要领域,作为差异性分析的主要工具——两样本检验一直是研究的重点和热点.但以往的大部分研究是围绕着欧氏数据展开的,而对于非欧氏数据的研究则是刚刚起步.本文尝试总结一些现有的非参数两样本检验的内在规律,由此构造新的两样本的非参数检验,可用于结构更为复杂的数据.此外,本文还讨论此研究方向所面临的挑战.
Abstract
The advancement of cutting-edge technology produces huge data sets.The current data boom not only poses a challenge to the traditional theory of statistical inference but also expands its scope.An important area of statistical inference research is difference analysis using various data types.The two-sample test is the primary tool for difference analysis,which has always been the focus and hot spot in the statistical community.However,previous research on the two-sample test has been focused on Euclidean data,whereas related studies on non-Euclidean data are still in their infancy.This study reviews some common characteristics of existing nonparametric two-sample tests.Then,based on this information,we create new two-sample nonparametric tests for data with more complex structures.In addition,the problems facing this field of study are discussed.
关键词
两样本检验/秩检验/最大平均差异/能量距离/球散度Key words
two-sample test/rank test/maximum mean discrepancy/energy distance/ball divergence引用本文复制引用
基金项目
国家自然科学基金重点项目(12231017)
国家自然科学基金面上项目(72171216)
出版年
2024
NETL
NSTL
万方数据