统计与决策2019，Vol.35Issue(8)：22-25,4.DOI:10.13546/j.cnki.tjyjc.2019.08.005
有限样本条件下重复抽样建模方法研究
Modeling Method of Repeated Sampling Under Finite Sample Condition
摘要
Abstract
This paper proves the superiority of the Features Sample Re-sampling Regression (FSR) method from the perspective of mathematics and simulation. Firstly the paper mathematically proves that for continuous variables, the sampling method in FSR can make full use of the data distribution information of non-original sampling points to improve the estimation accuracy, and that the statistical coefficient of variation obtained by FSR method is inversely proportional to the number of samples [m]. And then the paper makes simulation and comparison of small sample parameter estimation's traditional regression, Bayes Bootstrap and the method of features sample repeated sampling to verify the above conclusions.关键词
有限样本/特征样本重复抽样回归/先验分布Key words
finite sample/Features Sample Re-sampling Regression/prior distribution分类
数理科学引用本文复制引用
高艳平,王晶..有限样本条件下重复抽样建模方法研究[J].统计与决策,2019,35(8):22-25,4.基金项目
国家社会科学基金资助项目(16BTJ016) （16BTJ016）
