摘要
针对多元线性回归方程的显著性检验问题,在随机误差服从任意分布的假设下,运用线性模型和随机变量二次型的相关理论,推导出总离差平方和分解式中各分量的数学期望,进而从定量的角度说明回归方程显著性检验中所取拒绝域的合理性。
Issues of the significance test of multiple linear regression equation are discussed. For the random error obeying an arbitrary distribution, the expectations of components of the decomposition of total dispersion square sum are deduced based on the theories of linear model and quadratic form of random variable. Further, rejection region set in the significance test of multiple linear regression equation is theoretically proved reasonable.
出处
《绵阳师范学院学报》
2009年第11期8-11,共4页
Journal of Mianyang Teachers＇ College
基金
中国民航总局教育研究基金项目(CD0528)
关键词
多元线性回归
数学期望
显著性检验
multiple linear regression
expectation
significance test
