Density estimation with heteroscedastic error

被引:56
作者
Delaigle, Aurore [1 ,2 ]
Meister, Alexander [3 ]
机构
[1] Univ Bristol, Dept Math, Bristol BS8 1TW, Avon, England
[2] Univ Melbourne, Dept Math & Stat, Melbourne, Vic 3010, Australia
[3] Univ Ulm, D-89081 Ulm, Germany
关键词
bandwidth; density deconvolution; errors-in-variables; heteroscedastic contamination; inverse problems; plug-in;
D O I
10.3150/08-BEJ121
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
It is common, in deconvolution problems, to assume that the measurement errors are identically distributed. In many real-life applications, however, this condition is not satisfied and the deconvolution estimators developed for homoscedastic errors become inconsistent. In this paper, we introduce a kernel estimator of a density in the case of heteroscedastic contamination. We establish consistency of the estimator and show that it achieves optimal rates of convergence under quite general conditions. We study the limits of application of the procedure in some extreme situations, where we show that, in some cases, our estimator is consistent, even when the scaling parameter of the error is unbounded. We suggest a modified estimator for the problem where the distribution of the errors is unknown, but replicated observations are available. Finally, an adaptive procedure for selecting the smoothing parameter is proposed and its finite-sample properties are investigated on simulated examples.
引用
收藏
页码:562 / 579
页数:18
相关论文
共 33 条
[1]  
Bennett C.A., 1954, Statistical analysis in chemistry and the chemical industry
[2]   Minimax estimation of the noise level and of the deconvolution density in a semiparametric convolution model [J].
Butucea, C ;
Matias, C .
BERNOULLI, 2005, 11 (02) :309-340
[3]  
Carroll J., 2006, MEASUREMENT ERROR NO, V2nd edn
[4]   ROBUST LINEAR-REGRESSION IN REPLICATED MEASUREMENT ERROR MODELS [J].
CARROLL, RJ ;
ELTINGE, JL ;
RUPPERT, D .
STATISTICS & PROBABILITY LETTERS, 1993, 16 (03) :169-175
[5]   Low order approximations in deconvolution and regression with errors in variables [J].
Carroll, RJ ;
Hall, P .
JOURNAL OF THE ROYAL STATISTICAL SOCIETY SERIES B-STATISTICAL METHODOLOGY, 2004, 66 :31-46
[6]   OPTIMAL RATES OF CONVERGENCE FOR DECONVOLVING A DENSITY [J].
CARROLL, RJ ;
HALL, P .
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 1988, 83 (404) :1184-1186
[7]   Practical bandwidth selection in deconvolution kernel density estimation [J].
Delaigle, A ;
Gijbels, I .
COMPUTATIONAL STATISTICS & DATA ANALYSIS, 2004, 45 (02) :249-267
[8]  
DELAIGLE A, 2007, DENSITY ESTIMATION H
[9]  
Delaigle A, 2008, STAT SINICA, V18, P1025
[10]   On deconvolution with repeated measurements [J].
Delaigle, Aurore ;
Hall, Peter ;
Meister, Alexander .
ANNALS OF STATISTICS, 2008, 36 (02) :665-685