Asymptotic properties of local polynomial regression with missing data and correlated errors

被引:16
|
作者
Perez-Gonzalez, A. [1 ]
Vilar-Fernandez, J. M. [2 ]
Gonzalez-Manteiga, W. [3 ]
机构
[1] Univ Vigo, Fac Business Sci, Dept Stat & Operat Res, Orense, Spain
[2] Univ A Coruna, Fac Comp Sci, Dept Math, La Coruna, Spain
[3] Univ Santiago Compostela, Fac Math, Dept Stat & Operat Res, Santiago De Compostela, Spain
关键词
Local polynomial regression; Missing response and correlated errors; TIME-SERIES; NONPARAMETRIC REGRESSION; MODELS; AUTOREGRESSION; IMPUTATION; INFERENCE; VALUES;
D O I
10.1007/s10463-007-0136-2
中图分类号
O21 [概率论与数理统计]; C8 [统计学];
学科分类号
020208 ; 070103 ; 0714 ;
摘要
The main objective of this work is the nonparametric estimation of the regression function with correlated errors when observations are missing in the response variable. Two nonparametric estimators of the regression function are proposed. The asymptotic properties of these estimators are studied; expresions for the bias and the variance are obtained and the joint asymptotic normality is established. A simulation study is also included.
引用
收藏
页码:85 / 109
页数:25
相关论文
共 50 条