摘要
残差蕴含了线性回归模型基本假定的许多重要信息,文章通过分析残差,探讨回归模型基本假定的适应性和模型的合理性问题。以预测某一类型的肝手术病人的生存时间为例,利用残差的特点反推数据的适用性和模型的合理性。通过相应的措施实施改进,建立“最优”回归模型,减少盲目尝试修改模型的做法,为模型优化提供方向,为问题解决提供有效途径。
Residuals contain a lot of important information about the basic assumptions of linear regression models. By analyzing residuals, this paper discusses the adaptability of the basic assumptions of the regression model and the rationality of the model. The paper takes the prediction of the survival time of a certain type of liver surgery patients as an example, and uses the characteristics of residuals to reversely deduce the applicability of the data and the rationality of the model. Finally, Through the corresponding measures to implement the improvement, the paper establishes the optimal regression model, and reduces the blind attempt to modify the model, which provides the direction for model optimization and an effective way for problem solving.
作者
胡菊华
Hu Juhua(College of Science,Jiangxi Agricultural University,Nanchang 330045,China)
出处
《统计与决策》
CSSCI
北大核心
2019年第24期5-8,共4页
Statistics & Decision
关键词
残差分析
残差图
线性回归模型
residual analysis
residual plot
linear regression model
