摘要
半参数回归模型L=Bx+s+ε是线性模型与非参数回归模型的混合体。本文采用自然样条逼近方法,探讨了自然样条半参数回归分析方法,同时将此方法运用到形变分析与预报数据处理中,结果表明是一种有效的方法。半参数回归效果要优于参数回归,多项式参数回归的残差平方和远大于参数回归。
The semiparametric regression model L=Bx+s+ε is a combination of classical linear model and nonparametric regression model.The semiparametric regression method is discussed through natural spline fitting in this paper,at the same time the method is used to analyze and forecast deformation.The method proved to be effective from the results.
出处
《测绘科学》
CSCD
2004年第5期38-40,共3页
Science of Surveying and Mapping
基金
国家自然科学基金项目(460274005)
关键词
半参数回归
自然样条
曲线拟合
补偿最小二乘
模型误差
形变分析与预报
semiparametric regression
natural spline
curve fitting
penalized least squates
model errors
analysis and forecast of deformation
