×

Estimating a parameter when it is known that the parameter exceeds a given value. (English) Zbl 1337.62054

Summary: In some statistical problems a degree of explicit, prior information is available about the value taken by the parameter of interest, \(\theta\) say, although the information is much less than would be needed to place a prior density on the parameter’s distribution. Often the prior information takes the form of a simple bound, ‘\(\theta > \theta_{1}\)’ or ‘\(\theta < \theta_{1}\)’, where \(\theta_{1}\) is determined by physical considerations or mathematical theory, such as positivity of a variance. A conventional approach to accommodating the requirement that \(\theta > \theta_{1}\) is to replace an estimator, \(\tilde \theta\), of \(\theta\) by the maximum of \(\tilde \theta\) and \(\theta_{1}\). However, this technique is generally inadequate. For one thing, it does not respect the strictness of the inequality \(\theta > \theta_{1}\), which can be critical in interpreting results. For another, it produces an estimator that does not respond in a natural way to perturbations of the data. In this paper we suggest an alternative approach, in which bootstrap aggregation, or bagging, is used to overcome these difficulties. Bagging gives estimators that, when subjected to the constraint \(\theta > \theta_{1}\), strictly exceed \(\theta_{1}\) except in extreme settings in which the empirical evidence strongly contradicts the constraint. Bagging also reduces estimator variability in the important case for which \(\tilde \theta\) is close to \(\theta_{1}\), and more generally produces estimators that respect the constraint in a smooth, realistic fashion.

MSC:

62G05 Nonparametric estimation
62G09 Nonparametric statistical resampling methods
62G20 Asymptotic properties of nonparametric inference
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Borra, Performance evaluation of bagging and boosting in nonparametric regression, Metron 59 pp 141– (2001) · Zbl 1108.62312
[2] Borra, Improving nonparametric regression methods by bagging and boosting, Comput. Statist. Data Anal. 38 pp 407– (2002) · Zbl 1072.62562
[3] Breiman, Bagging predictors, Machine Learning 24 pp 123– (1996) · Zbl 0858.68080
[4] BÃ{\(\tfrac14\)}hlmann, Handbook of Computational Statistics pp 877– (2004)
[5] BÃ{\(\tfrac14\)}hlmann, Analyzing bagging, Annals Statist. 30 pp 927– (2002)
[6] Buja, Observations on bagging, Statist. Sinica 16 pp 323– (2006)
[7] Chen, Effects of bagging and bias correction on estimators defined by estimating equations, Statist. Sinica 13 pp 97– (2003) · Zbl 1022.62020
[8] Dunn, Statistical Evaluation of Measurement Errors : Design and Analysis of Reliability Studies (2004)
[9] Friedman, On bagging and nonlinear estimation, J. Statist. Plann. Inf. 137 pp 669– (2007) · Zbl 1104.62047
[10] Gibbons, Weighted random-effects regression models with application to interlaboratory calibration, Technometrics 43 pp 192– (2001)
[11] Hall, Reducing variability of cross-validation for smoothing parameter choice, Biometrika 96 pp 175– (2009)
[12] Hall, Properties of bagged nearest neighbour classifiers, J. R. Statist. Soc. Ser. B Statist. 67 pp 363– (2005) · Zbl 1069.62051
[13] Hothorn, Bundling classifiers by bagging trees, Comput. Statist. Data Anal. 49 pp 1068– (2005) · Zbl 1429.62246
[14] Ridgeway, Looking for lumps: boosting and bagging for density estimation, Comput. Statist. Data Anal. 38 pp 379– (2002) · Zbl 1072.62560
[15] Sinisi, Cross-validated bagged prediction of survival, Statist. Appl. Genet. Mol. Biol. 5 pp 26– (2006) · Zbl 1166.62326
[16] Snee, Graphical display of results of three-treatment randomized block experiments, Appl. Statist. 34 pp 71– (1985)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.