自然样条半参数模型与系统误差估计

基金项目: 国家自然科学基金资助项目(460274005)

详细信息

Semiparametric Regression Model With Natural Spline and Systematic Error Estimation

School of Geodesy and Geomatics, Wuhan University, 129 Luoyu Road, Wuhan 430079, China

摘要: 采用自然样条逼近的数据处理方法,探讨了自然样条半参数回归分析方法。在补偿最小二乘的原则下,利用三次样条函数构造补偿项,通过广义交叉核实函数自动选取光滑参数。自编程序进行计算,得到了回归参数向量和样条函数的补偿最小二乘估计。模拟计算表明,该方法适合于回归函数模型误差与测量系统误差的估计。
- 半参数回归 /
- 自然样条 /
- 曲线拟合 /
- 补偿最小二乘 /
- 系统误差估计
Abstract: The semiparametric regression is discussed by natural spline fitting. According to penalized least squares, Cubic natural spline forms penalized item. The choice of smoothing parameter can be automatically obtained by using generalized cross-validation function. Estimators of penalized least square for parametric regression and vectors and spline function can be got by compiled program.
- semiparametric regression /
- natural spline /
- curve fitting /
- penalized least squares /
- systematic error estimation