摘要
半参数模型保留了参数模型和非参数模型好的性质,具有广泛的实用价值,故对半参数模型的研究具有重要意义,是目前研究的热点问题。笔者主要采用惩罚最小二乘方法,给出了对带非参平滑项的半参模型在线性部分受约束条件下的一种新的估计方法,并且得到在平滑矩阵S为对称及任意两种情况下的估计,及新估计量的协方差.并对比了新方法与传统方法的异同.
The semiparametric model retains the good-nature of parameter and nonparameter models, with a wide range of practical value, so studying the semiparametric model is of great significance. It is currently a hot issue. By means of the penalized least-squares method, this paper presents a method of estimation of treatment effects in a semiparametric model with one smoothing term under additional conditions on their linear functions {Rβ=d^yn×1=Xn×ρβρ×1＋fn×1＋en×1 and obtains estimates in both cases of smooth matrix S for symmetry and arbitrary ,and new estimates of covariance. In conclusion, it compares the similarities and differences of the new method and the traditional method.
出处
《青岛理工大学学报》
CAS
2008年第4期117-119,共3页
Journal of Qingdao University of Technology
关键词
半参回归模型
惩罚最小二乘
平滑估计
向后算法
semiparametric regression model
penalized least-squares
smoothing estimate
backfitting algorithm
