Abstract
In this paper, the problem of bandwidth choice in smooth k-sample tests is considered. Three different bootstrap methods are discussed and implemented. All the methods persecute the bandwidth leading to the maximum power, while preserving the level of the test. The relative performance of the methods is investigated in a simulation study. Illustration through real medical data is provided. The main conclusion is that the bootstrap minimum method provides a good compromise between statistical power and conservativeness. Robustness of the methods with respect to the number of bootstrap resamples and practical limitations are discussed.
Similar content being viewed by others
References
Ahmad AI, Amezziane M (2007) A general and fast convergent bandwidth selection method of kernel estimator. J Nonparametr Stat 19:165–187
Anderson NH, Hall P, Titterington DM (1994) Two-sample test statistics for measuring discrepancies between two multivariate probability density functions using kernel-based density estimates. J Multivar Anal 50:41–54
Bowman A, Azzalini A (2001) Applied smoothing techniques for data analysis. Oxford University Press, Oxford
Breiman L (1996) Bagging predictors. Mach Learn 24:123–140
Cao R, Lugosi I (2005) Goodness-of-fit based on the kernel density estimator. Scand J Stat Theory Appl 32:599–615
Cao R, Van Keilegom I (2006) Empirical likelihood tests for two-sample problems via nonparametric density estimation. Canad J Stat 34:61–77
Devroye L (1997) Universal smoothing factor selection in density estimation: theory and practice. Test 6(2):223–320
Gao J, Gijbels I (2008) Bandwidth selection in nonparametric kernel testing. J Am Stat Assoc 484:1584–1594
González JR, Carrasco JL, Dudbridge F, Estivill X, Moreno V (2008) Maximizing association statistics over genetic models. Genet Epidemiol 32:246–254
Hall P, DiCiccio JT, Romano JP (1989) On smoothing and the bootstrap. Ann Stat 17(2):692–704
Hart JD (1997) Nonparametric smoothing and lack-of-fit tests. Springer, New York
Horváth L, Steinebach J (1999) On the best approximation for bootstrapped empirical processes. Stat Probab Lett 41:117–122
Louani D (1998) Large deviations limit theorems for the kernel density estimator. Scand J Stat 25:243–253
Louani D (2000) Large deviation for \(L_1\)-distance in kernel density estimation. J Stat Plan Inference 90:177–182
Martínez-Camblor P (2010) Nonparametric \(k\)-sample tests based on kernel density estimator for paired design. Comput Stat Data Anal 54:2035–2045
Martínez-Camblor P, de Uña-Álvarez J (2009) Nonparametric \(k\)-sample tests: density function vs. distribution function. Comput Stat Data Anal 53(9):3344–3357
Martínez-Camblor P, de Uña-Álvarez J (2012) Density comparison for independent and right censored data via kernel density estimator. Comput Stat. doi:10.1007/s00180-011-0298-5
Martínez-Camblor P, de Uña-Álvarez J, Corral N (2008) \(k\)-Sample test based on the common area of kernel density estimator. J Stat Plan Inference 138(12):4006–4020
Martínez-Camblor P, Larrañaga N, Sarasqueta C, Basterretxea M (2009) “Esa corporeidad mortal y rosa”: Análisis del tiempo libre de enferemedad del cáncer de mama en Gipuzkoa en presencia de riesgos competitivos. Gaceta Sanitaria 23(6):554–447
Park BU, Marron JS (1990) Comparison of data-dirven bandwidth selectors. J Am Stat Assoc 85(409):66–72
Zhang J, Wu Y (2007) \(k\)-Sample test based on the likelihood ratio. Comput Stat Data Anal 51(9):4682–4691
Acknowledgments
The authors are grateful to Nerea Larrañaga (Subdirección de Salud Pública de Gipuzkoa) and her colleagues for permission to use the breast cancer data. This work was supported by the Grant MTM2011-23204 of the Spanish Ministry of Science and Innovation (FEDER support included. Second author was also supported by the research Grant MTM2008-03129 of the Spanish Ministerio de Ciencia e Innovación, and by the Grant 10PXIB300068PR of the Xunta de Galicia.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Martínez-Camblor, P., de Uña-Álvarez, J. Studying the bandwidth in \(k\)-sample smooth tests. Comput Stat 28, 875–892 (2013). https://doi.org/10.1007/s00180-012-0333-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00180-012-0333-1