A Bayesian approach to hybrid splines non-parametric regression

被引：6

作者：

Dias, R ^{[1
]}

Gamerman, D

机构：

[1] Univ Estadual Campinas, Dept Estatist, Sao Paulo, Brazil

[2] Univ Fed Rio de Janeiro, Inst Matemat, Rio De Janeiro, Brazil

来源：

JOURNAL OF STATISTICAL COMPUTATION AND SIMULATION | 2002年 / 72卷 / 04期

基金：

巴西圣保罗研究基金会;

关键词：

hybrid splines non-parametric regression; reversible jump MCMC;

D O I：

10.1080/00949650212844

中图分类号：

TP39 [计算机的应用];

学科分类号：

081203 ; 0835 ;

摘要：

A Bayesian approach is considered to estimate the number of basis functions and the smoothing parameter of the hybrid splines non-parametric regression procedure. The method used to obtain the estimate of the regression curve and its Bayesian confidence intervals is based on the reversible jump MCMC (Green 1995). Illustrations with simulated data are provided and show good performance of the proposed approach over the existing methods.

引用

页码：285 / 297

页数：13

共 19 条

[1] Improved inference in nonparametric regression using L(k)-smoothing splines
Abramovich, F
Steinberg, DM
[J]. JOURNAL OF STATISTICAL PLANNING AND INFERENCE, 1996, 49 (03) : 327 - 341
[2] BATES D, 1982, COMPUTATIONAL METHOD
[3] SMOOTHING NOISY DATA WITH SPLINE FUNCTIONS
WAHBA, G
[J]. NUMERISCHE MATHEMATIK, 1975, 24 (05) : 383 - 393
[4] de Boor C., 1978, PRACTICAL GUIDE SPLI, DOI DOI 10.1007/978-1-4612-6333-3
[5] Automatic Bayesian curve fitting
Denison, DGT
Mallick, BK
Smith, AFM
[J]. JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY, 1998, 60 : 333 - 350
[6] Density estimation via hybrid splines
Dias, R
[J]. JOURNAL OF STATISTICAL COMPUTATION AND SIMULATION, 1998, 60 (04) : 277 - 293
[7] Sequential adaptive nonparametric regression via H-splines
Dias, R
[J]. COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION, 1999, 28 (02) : 501 - 515
[8] IDEAL SPATIAL ADAPTATION BY WAVELET SHRINKAGE
DONOHO, DL
JOHNSTONE, IM
[J]. BIOMETRIKA, 1994, 81 (03) : 425 - 455
[9] Gamerman D., 1997, MARKOV CHAIN MONTE C
[10] Green PJ, 1995, BIOMETRIKA, V82, P711, DOI 10.2307/2337340

← 1 2 →