Abstract
For the standard kernel density estimate, it is known that one can tune the bandwidth such that the expected Li error is within a constant factor of the optimal L1 error (obtained when one is allowed to choose the bandwidth with knowledge of the density). In this paper, we pose the same problem for variable bandwidth kernel estimates where the bandwidths are allowed to depend upon the location. We show in particular that for positive kernels on the real line, for any data-based bandwidth, there exists a density for which the ratio of expected L1 error over optimal L1 error tends to infinity. Thus, the problem of tuning the variable bandwidth in an optimal manner is “too hard”. Moreover, from the class of counterexamples exhibited in the paper, it appears that placing conditions on the densities (monotonicity, convexity, smoothness) does not help.
Research was supported by NSERC Grant A3456 and by FCAR Grant 90-ER-0291.
Research was supported by DIGES Grant PB96-0300.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
I. Abramson, On bandwidth variation in kernel estimates—a square root law, Annals of Statistics, vol. 10, pp. 1217–1223, 1982.
H. Akaike, An approximation to the density function, Annals of the Institute of Statistical Mathematics, vol. 6, pp. 127–132, 1954.
P. Assouad, Deux remarques sur l’estimation, Comptes Rendus de l’Académie des Sciences de Paris, vol. 296, pp. 1021–1024, 1983.
A. P. Basu, Estimates of reliability for some distributions useful in life testing, Technometrics, vol. 6, pp. 215–219, 1964.
L. Breiman, W. Meisel, and E. Purcell, Variable kernel estimates of multivariate densities, Technometrics, vol. 19, pp. 135–144, 1977.
L. Birgé, Non-asymptotic minimax risk for Hellinger balls, Probability and Mathematical Statistics, vol. 5, pp. 21–29, 1985.
L. Birgé, On estimating a density using Hellinger distance and some other strange facts, Probability Theory and Related Fields, vol. 71, pp. 271–291, 1986.
L. Birgé, On the risk of histograms for estimating decreasing densities, Annals of Statistics, vol. 15, pp. 1013–1022, 1987a.
L. Birgé, Estimating a density under order restrictions: nonasymptotic minimax risk, Annals of Statistics, vol. 15, pp. 995–1012, 1987b.
L. Birgé, The Grenander estimator: a nonasymptotic approach, Annals of Statistics, vol. 17, pp. 1532–1549, 1989.
L. Devroye, A note on the Ll consistency of variable kernel estimates, Annals of Statistics, vol. 13, pp. 1041–1049, 1985.
L. Devroye, A Course in Density Estimation, Birkhäuser, Boston, 1987.
L. Devroye, Another proof of a slow convergence result of Birgé, Statistics and Probability Letters, vol. 23, pp. 63–67, 1995.
L. Devroye, Universal smoothing factor selection in density estimation: theory and practice, Test, vol. 6, pp. 223–320, 1997.
L. Devroye and L. Györfi, Nonparametric Density Estimation: The L1 View, John Wiley, New York, 1985.
L. Devroye and G. Lugosi, A universally acceptable smoothing factor for kernel density estimation, Annals of Statistics, vol. 24, pp. 2499–2512, 1996.
L. Devroye and G. Lugosi, Non-asymptotic universal smoothing factors, kernel complexity and Yatracos classes, Annals of Statistics, vol. 25, pp. 2626–2637, 1997.
L. Devroye, G. Lugosi, and F. Udina, Inequalities for a new data-based method for selecting nonparametric density estimates, Technical Report, Facultat de Ciencies Economiques, Universitat Pompeu Fabra, Barcelona, 1998.
L. Devroye and C. S. Penrod, The strong uniform convergence of multivariate variable kernel estimates, Canadian Journal of Statistics, vol. 14, pp. 211–219, 1986.
M. Farmen, The smoothed bootstrap for variable bandwidth selection and some results in nonparametric logistic regression, Ph.D. Dissertation, Department of Statistics, University of North Carolina, Chapel Hill, 1996.
H. Guttmann and W. Wertz, Note on estimating normal densities, Sankhya, Series B, vol. 38, pp. 231–236, 1976.
J. D. F. Habbema, J. Hermans, and J. Remme, Variable kernel density estimation in discriminant analysis, in: COMPSTAT 1978: Proceedings, ed. L. C. A. Corsten and J. Hermans, Birkhauser, Basel, 1978.
P. Hall, On global properties of variable bandwidth density estimators, Annals of Statistics, vol. 20, pp. 762–778, 1992.
P. Hall and J. S. Marron, Variable window width kernel estimates, Probability Theory and related Fields, vol. 80, pp. 37–49, 1988.
P. Hall and W. R. Schucany, A local cross-validation algorithm, Statistics and Probability Letters, vol. 8, pp. 109–117, 1989.
M. Hazelton, Bandwidth selection for local density estimation, Scandinavian Journal of Statistics, vol. 23, pp. 221–232, 1996.
M. C. Jones, Variable kernel density estimates and variable kernel density estimates, Australian Journal of Statistics, vol. 32, pp. 361–371, 1990.
Ya. P. Lumelskii and P. N. Sapozhnikov, Unbiased estimates of density functions, Theory of Probability and its Applications, vol. 14, pp. 357–364, 1969.
J. S. Marron, P. Hall, and T. C. Hu, Improved variable window estimators of probability densities, Annals of Statistics, vol. 23, pp. 1–10, 1995.
J. Mielniczuk, P. Sarda, and P. Vieu, Local data-driven bandwidth choice for density estimation, Journal of Statistical Planning and Inference, vol. 23, pp. 53–69, 1989.
E. Parzen, On the estimation of a probability density function and the mode, Annals of Mathematical Statistics, vol. 33, pp. 1065–1076, 1962.
J. W. Raatgever and R. P. W. Duin, On the variable kernel model for multivariate nonparametric density estimation, in: COMPSTAT 1978: Proceedings, ed. L. C. A. Corsten and J. Hermans, Birkhauser, Basel, 1978.
M. Rosenblatt, Remarks on some nonparametric estimates of a density function, Annals of Mathematical Statistics, vol. 27, pp. 832–837, 1956.
S. R. Sain, Adaptive kernel density estimation, Ph.D. Dissertation, Department of Statistics, Rice University, Houston, 1994.
S. R. Sain and D. W. Scott, On locally adaptive density estimation, Journal of the American Statistical Association, vol. 91, pp. 1525–1534, 1996.
S. R. Sain and D. W. Scott, Zero-bias locally adaptive density estimators, Technical Report, Rice University, Houston, 1997.
A. H. Seheult and C. P. Quesenberry, On unbiased estimation of density functions, Annals of Mathematical Statistics, vol. 42, pp. 1434–1438, 1971.
S. J. Sheather, A data-based algorithm for choosing the window width when estimating the density at a point, Computational Statistics and Data Analysis, vol. 1, pp. 229–239, 1983.
S. J. Sheather, An improved data-based algorithm for choosing the window width when estimating the density at a point, Computational Statistics and Data Analysis, vol. 4, pp. 61–65, 1986.
G. R. Terrell and D. W. Scott, Variable kernel density estimation, Annals of Statistics, vol. 20, pp. 1236–1265, 1992.
L. A. Thombs and S. J. Sheather, Local bandwidth selection for density estimation, in: Proceedings of the 22nd Symposium on the Interface, pp. 111–116, Springer-Verlag, New York, 1992.
W. Wertz, On unbiased density estimation, An. Acad. Brasil. Cienc. vol. 47, pp. 65–72, 1975.
Y. Yang, Minimax Optimal Density Estimation, Ph.D. Dissertation, Yale University, 1996.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2000 Springer Science+Business Media New York
About this paper
Cite this paper
Devroye, L., Lugosi, G. (2000). Variable Kernel Estimates: on the Impossibility of Tuning the Parameters. In: Giné, E., Mason, D.M., Wellner, J.A. (eds) High Dimensional Probability II. Progress in Probability, vol 47. Birkhäuser, Boston, MA. https://doi.org/10.1007/978-1-4612-1358-1_27
Download citation
DOI: https://doi.org/10.1007/978-1-4612-1358-1_27
Publisher Name: Birkhäuser, Boston, MA
Print ISBN: 978-1-4612-7111-6
Online ISBN: 978-1-4612-1358-1
eBook Packages: Springer Book Archive