基于大数据思维的统计学若干理论问题

Several Theoretical Problems in Statistics Based on the Idea of Big Data

本文基于大数据思维,对统计学若干理论问题进行了讨论。本文认为:在大数据背景下,统计学以数据为研究对象没有变,作为方法论科学的性质没有变,以探索真相为根本任务没有变;但是作为统计学研究对象的数据出现了四个新的动向,即数据内涵有新理解,数据结构与来源有新方式,数据不确定性有新表现,数据分析思路与技术有新要求;为此,需要重新理解和定义总体、个体与样本,标志与变量,可靠性与有效性这三组统计学中最基本也最重要的概念。最后本文指出,大数据为统计学发展提供了新的平台,需要我们从数据、方法和应用三个维度同时发力。

This paper discusses several theoretical problems in statistics based on the idea of big data. It argues that within the framework of big data, data remains as the research object of statistics, statistics remains as a methodological science and approaching the truth remains as the essential goal. There are emergingfour new trends in data as the object： new interpretations of its meaning, new methods to source and structure it, new indications of its uncertainty and new requirement of the logic and technology. Therefore, it is necessary to re-interpret and redefine the following three groups of fundamental concepts that are of most significance： population, individual and sample; indication and variable; reliability and effectiveness. Finally, this paper holds that big data has provided a new platform for the development of statistics and should be studied simultaneously in the dimensions of data, methodology and applieation.