This paper introduces a computationally tractable density estimator that has the same asymptotic variance as the classical Nadaraya-Watson density estimator but whose asymptotic bias is zero. We achieve this result using a two stage estimator that applies a multiplicative bias correction to an oversmooth pilot estimator. Simulations show that our asymptotic results are available for samples as low as , where we see an improvement of as much as 20% over the traditionnal estimator.
Mots clés : nonparametric density estimation, kernel smoother, asymptotic normality, bias reduction, confidence intervals
@article{PS_2009__13__1_0, author = {Hengartner, Nicolas W. and Matzner-L{\o}ber, \'Eric}, title = {Asymptotic unbiased density estimators}, journal = {ESAIM: Probability and Statistics}, pages = {1--14}, publisher = {EDP-Sciences}, volume = {13}, year = {2009}, doi = {10.1051/ps:2007055}, mrnumber = {2493852}, language = {en}, url = {http://www.numdam.org/articles/10.1051/ps:2007055/} }
TY - JOUR AU - Hengartner, Nicolas W. AU - Matzner-Løber, Éric TI - Asymptotic unbiased density estimators JO - ESAIM: Probability and Statistics PY - 2009 SP - 1 EP - 14 VL - 13 PB - EDP-Sciences UR - http://www.numdam.org/articles/10.1051/ps:2007055/ DO - 10.1051/ps:2007055 LA - en ID - PS_2009__13__1_0 ER -
Hengartner, Nicolas W.; Matzner-Løber, Éric. Asymptotic unbiased density estimators. ESAIM: Probability and Statistics, Tome 13 (2009), pp. 1-14. doi : 10.1051/ps:2007055. http://www.numdam.org/articles/10.1051/ps:2007055/
[1] On bandwidth variation in kernel estimates - a square root law. Ann. Statist. 10 (1982) 1217-1223. | MR | Zbl
,[2] Adaptive density flattening-metric distortion principle for combining bias in nearest neighbor methods. Ann. Statist. 12 (1984) 880-886. | MR | Zbl
,[3] Distribution Estimation Consistent in Total Variation and in Two Types of Information Divergence. IEEE Trans. Inf. Theory 38 (1992) 1437-1453. | MR | Zbl
, and ,[4] Hierarchies of higher order kernels. Prob. Theory Related Fields 94 (1993) 489-504. | MR | Zbl
,[5] Optimizing kernel methods: a unifying variational principle. Ins. Statist. Rev. 59 (1991) 373-388. | Zbl
and ,[6] On the bias of variable bandwidth curve estimators. Biometrika 77 (1990) 529-535. | MR | Zbl
,[7] Variable window width kernel estimates of a probability density. Prob. Theory Related Fields 80 (1988) 37-49. | MR | Zbl
and ,[8] Nonparametric density estimation with a parametric start. Ann. Statist. 23 (1995) 882-904. | MR | Zbl
and ,[9] Locally parametric nonparametric density estimation. Ann. Statist. 24 (1996) 1619-1647. | MR | Zbl
and ,[10] Variable kernel density estimates variable kernel density estimates. Aust. J. Statist. 32 (1990) 361-371. Correction 33 (1991) 119. | MR
,[11] A simple bias reduction method for density estimation. Biometrika 82 (1995) 327-38. | MR | Zbl
, and ,[12] Variable location and scale kernel density estimation. Inst. Statist. Math. 46 (1994) 521-535. | MR | Zbl
, and ,[13] A note on bias reduction in variable kernel density estimates. Can. J. Statist. 21 (1993) 367-375. | MR | Zbl
,[14] Exact mean integrated squared error. Ann. Statist. 20 (1992) 712-736. | MR | Zbl
and ,[15] A multiplicative bias reduction method for nonparametric regression. Statist. Probab. Lett. 19 (1994) 181-187. | MR | Zbl
and ,[16] Remarks on some nonparametric estimates of a density function. Ann. Math. Statist. 27 (1956) 832-837. | MR | Zbl
,[17] A law of iterated logarithm for kernel density estimators. Ann. Probab. 10 (1982) 414-422. | MR | Zbl
,[18] On improving convergence rates for non-negative kernel density estimators. Ann. Statist. 8 (1980) 1160-1163. | MR | Zbl
and ,[19] Kernel Smoothing. Chapman and Hall, London (1995). | MR | Zbl
and ,Cité par Sources :