Abstract.
Based on a random sample of size n from an unknown d-dimensional density f, the problem of selecting the bandwidths in kernel estimation of f is investigated. The optimal root n relative convergence rate for bandwidth selection is established and the information bounds in this convergence are given, and a stabilized bandwidth selector (SBS) is proposed. It is known that for all d the bandwidths selected by the least squares cross-validation (LSCV) have large sample variations. The proposed SBS, as an improvement of LSCV, will reduce the variation of LSCV without significantly inflating its bias. The key idea of the SBS is to modify the d-dimensional sample characteristic function beyond some cut-off frequency in estimating the integrated squared bias. It is shown that for all d and sufficiently smooth f and kernel, if the bandwidth in each coordinate direction varies freely, then the multivariate SBS is asymptotically normal with the optimal root n relative convergence rate and achieves the (conjectured) ‘‘lower bound’’ on the covariance matrix.
References
Brillinger, D. R.: Time Series Data Analysis and Theory. Holt, Rinehart and Winston, New York, (1981)
Chiu, S. T.: Bandwidth selection for kernel density estimation. Ann. Statist. 19, 1883–1905 (1991a)
Chiu, S. T.: The effect of discretization error on bandwidth selection for kernel density estimation. Biometrika 78, 436–441 (1991b)
Chiu, S. T.: An automatic bandwidth selector for kernel density estimation. Biometrika 79, 771–782 (1992)
Fan, J., Marron, J. S.: Best possible constant for bandwidth selection. Ann. Statist. 20, 2057–2070 (1992)
Fukunaga, K.: Introduction to Statistical Pattern Recognition. Academic Press, NY, 1972
Hall, P., Marron, J. S.: Lower bounds for bandwidth selection in density estimation. Probab. Theo. Rel. Fields 90, 149–173 (1991a)
Hall, P., Marron, J. S.: Local minimum in cross-validations. J. R. Statist. Soc. B 53, 245–252 (1991b)
Hall, P., Marron, J. S., Park, B. U.: Smoothed cross-validation. Probab. Theo. Rel. Fields 92, 1–20 (1992)
Jones, M. C.: The role of ISE and MISE in density estimation. Statist. and Probab. Letters 12, 51–56 (1991)
Jones, M. C., Marron, J. S., Park, B. U.: A simple root n bandwidth selector. Ann. Statist. 19, 1919–1932 (1991)
Jones, M. C., Marron, J. S., Sheather, S. J.: A Brief Survey of Bandwidth Selection for Density Estimation. J. Amer. Statist. Assoc. 91, 401–407 (1996)
Marron, J. S.: Will the art of smoothing ever become a science? Contemp. Math. 59, 169–178 (1986)
Marron, J. S.: Automatic smoothing parameter selection: a survey. Empirical Economics 13, 187–208 (1988)
Park, B. U., Marron, J. S.: Comparison of data-driven bandwidth selectors. J. Amer. Statist. Assoc. 85, 66–78 (1990)
Park, B. U., Marron, J. S.: On the use of pilot estimators in bandwidth selection. J. Nonparametric Statist. 1, 231–240 (1992)
Sain, S. R., Baggerly, K. A., Scott, D.W.: Cross-Validation of Multivariate Densities. J. Amer. Statist. Assoc. 89, 807–817 (1994)
Scott, D. W.: Multivariate Density Estimation: Theory, Practice, and Visualization. John Wiley, New York, (1992)
Scott, D. W., Terrell, G. R.: Biased and unbiased cross-validation in density estimation. J. Amer. Statist. Assoc. 82, 1131–1146 (1987)
Sheather, S. J., Jones, M. C.: A reliable data-based bandwidth selection method for kernel density estimation. J. R. Statist. Soc. B 53, 683–690 (1991)
Silverman, B. W.: Density Estimation for Statistics and Data Analysis. Chapman and Hall, London, (1986)
Stone, C. J.: Optimal convergence rates for nonparametric estimators. Ann. Statist. 8, 1348–1360 (1980)
Stone, C. J.: An asymptotically optimal window selection rule for kernel density estimates. Ann. Statist. 12, 1285–1297, (1984)
Taylor, C.C.: Bootstrap Choice of the Smoothing Parameter in Kernel Density Estimates. Biometrika 76, 705–712 (1989)
Terrell, G.R.: The maximal smoothing principle in density estimation. J. Amer. Statist. Assoc. 85, 470–477 (1990)
Terrell, G.R., Scott, D. W.: Oversmoothed Nonparametric Density Estimates. J. Amer. Statist. Assoc. 80, 209–214 (1985)
Wand, M. P.: Error Analysis for general multivariate kernel estimators. Nonparametric Statist. 2, 1–15 (1992)
Wand, M.P., Jones, M. C.: Comparison of Smoothing Parameterizations in Bivariate Kernel Density Estimation. J. Amer. Statist. Assoc. 88, 520–528 (1993)
Wand, M.P., Jones, M. C.: Multivariate Plug-in Bandwidth Selection. Computational Statist. 9, 97–116 (1994)
Worton, B. J.: Optimal Smoothing Parameters for Multivariate Fixed and Adaptive Kernel Methods. J. Statist. Comput. Simul. 32, 45–57 (1989)
Wu, T.-J.: Adaptive root n estimates of integrated squared density derivatives. Ann. Statist. 23, 1474–1495 (1995)
Wu, T.-J.: Root n bandwidth selectors for kernel estimation of density derivatives. J. Amer. Statist. Assoc. 92, 536–547 (1997)
Wu, T.-J., Lin, Y.: Information Bound for bandwidth selection in kernel estimation of density derivatives. Statistica Sinica 10, 457–473 (2000)
Wu, T.-J., Tsai, M.-H.: Root n bandwidths selectors in multivariate kernel density estimation (longer version). (2004) http://www.stat.ncku.edu.tw/tjwu/
Yang, L., Tschernig, R.: Multivariate bandwidth selection for local linear regression. J. R. Statist. Soc. B 61, 793–815 (1999)
Author information
Authors and Affiliations
Additional information
Part of the research was done while the first author was visiting the Institute of Statistical Science, Academia Sinica, Taipei, Taiwan. This work was supported by grant NSC-89-2118-M-006-011, NSC-90-2118-M-006-013 and NSC-91-2118-M-006-005 of National Science Council of Taiwan, R.O.C.
Rights and permissions
About this article
Cite this article
Wu, TJ., Tsai, MH. Root n bandwidths selectors in multivariate kernel density estimation. Probab. Theory Relat. Fields 129, 537–558 (2004). https://doi.org/10.1007/s00440-004-0357-8
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00440-004-0357-8
- Key words or phrases:
- Characteristic function
- Cross-validation
- Information bound
- Multivariate data
- Relative convergence rate