回归分析中最小样本容量的确定  

The Determination of the Minimum Sample Size in the Regression Analysis

作　　者：马赞甫[1] 刘妍珺[2] Ma Zanfu Liu Yanjun(guizhou key laboratory of economics system simulation b.school of mathematics and statics, Guizhou University of Finance and Economics, Guiyang 550025, China)

机构地区：[1]贵州财经大学贵州省经济系统仿真重点实验室,贵阳550025 [2]贵州财经大学数学与统计学院,贵阳550025

出　　处：《统计与决策》2017年第5期20-22,共3页Statistics & Decision

基　　金：教育部人文社会科学基金资助项目(12YJC790140)

摘　　要：在给定解释变量个数的情况下,样本容量影响到回归方程的稳定性。考虑以残差平方和的相对稳定表征样本回归方程的稳定性,借此构造一个服从贝塔分布或F分布的统计量以确定最小样本容量。In the case of a given number of explanatory variables, the sample size affects the stability of the regression equation. Considering that stability of the regression equation can be explained by the relative stability of residual sum of squares, which provides us with a good theoretical foundation to determine the minimum sample size by means of constructing a statistic obeying Beta distribution or F distribution.

关 键 词：回归分析 样本容量 残差平方和 贝塔分布 F分布 

分 类 号：O21[理学—概率论与数理统计]