摘要
利用线性回归建模时,一般都采用有常数项的模型,而不去考虑它对回归的作用是否显著。对于一元线性回归方程,在理论上对常数项显著性的判断进行了探讨;认为当其检验不显著的时候,重新建立新的回归方程,预测效果会更好;并用实际例子做了说明。
There is a regression constant in the linear regression model. But it is not significant in some conditions. Basing on the simple linear regression, how to judge the important of the regression constant is discussed. In the practical application, such as forecast, the regression model without constant is more power when the constant is not significant. A case to show this is given.
出处
《科学技术与工程》
2007年第8期1715-1716,1739,共3页
Science Technology and Engineering
关键词
线性回归
方差分析
显著性水平
linear regression analysis of variance signifcance level