Abstract
Let X be a real random variable having f as density function. Let F be its cumulative distribution function and Q its quantile function. For h > 0, let Fh and Qh denote respectively the Nadaraya kernel estimator of F and Q. In the first part of this paper the almost sure convergence of the conventional L1 distance between Qh and Q is established. In the second part, the L1 right inversion distance is introduced. The representation of this L1 right inversion distance in terms of Fh and F is given. This representation allows us to suggest ways to choose a global bandwidth for the estimator Qh.
Similar content being viewed by others
References
Adamowski, K. (1985). Nonparametric kernel estimation of flood frequencies. Water. Resour. Res. 21, 11, 1585–1590.
Altman, N. and Léger, C. (1995). Bandwidth selection for kernel distribution function estimation. J. Statist. Plann. Inference 46, 2, 195–214.
Apostol, T. M. (1974). Mathematical analysis, 2nd edn. Addison-Wesley, Reading. Mass.-London-Don Mills, Ont.
Azzalini, A. (1981). A note on the estimation of a distribution function and quantiles by a kernel method. Biometrika 68, 1, 326–328.
Bickel, P. J. and Freedman, D. A. (1981). Some asymptotic theory for the bootstrap. Ann. Statist. 9, 6, 1196–1217.
Bowman, A., Hall, P. and Prvan, T. (1998). Bandwidth selection for the smoothing of distribution functions. Biometrika 85, 4, 799–808.
Cai, Z. and Roussas, G. G. (1997). Smooth estimate of quantiles under association. Stat. Probab. Lett. 36, 3, 275–287.
Chacón, J. E. and Rodríguez-Casal, A. (2010). A note on the universal consistency of the kernel distribution function estimator. Stat. Probab. Lett. 80, 17-18, 1414–1419.
Cheng, M. -Y. and Sun, S. (2003). Bandwidth selection for kernel quantile estimation. unpublished.
Falk, M. (1985). Asymptotic normality of the kernel quantile estimator. Ann. Statist. 13, 1, 428–433.
Faucher, D., Rasmussen, P. F. and Bobée, B. (2001). A distribution function based bandwidth selection method for kernel quantile estimation. J. Hydrol.250, 1, 1–11.
Lejeune, M. and Sarda, P. (1992). Smooth estimators of distribution and density functions. Comput. Stat. Data Anal. 14, 4, 457–471.
Quintela-del Río, A. (2011). On bandwidth selection for nonparametric estimation in flood frequency analysis. Hydrol. Process. 25, 5, 671–678.
Sarda, P. (1993). Smoothing parameter selection for smooth distribution functions. J. Statist. Plann. Inference 35, 1, 65–75.
Schmid, F. (1993). Measuring interdistributional inequality. Springer, Technical report.
Shankar, B. (1998). An optimal choice of bandwidth for perturbed sample quantiles. Master’s thesis.
Shorack, G. R. and Wellner, J. A. (1986). Empirical processes with applications to statistics. Wiley, New York,.
Wang, L. (2008). Kernel type smoothed quantile estimation under long memory. Stat. Pap. 51, 1, 57–67.
Yamato, H. (1973). Some statistical properties of estimators of density and distribution functions. Bull. Math. Statist. 15, 1-2, 113–131.
Youndjé, E. (2018). Equivalence between mallows and girone–cifarelli tests. Stat. Probab. Lett. 141, 125–128.
Acknowledgments
My heartfelt gratitude goes to Professor William Strawderman of Rutgers University who helped me to improve the English level of this paper. I also greatly appreciated helpful comments from the referees.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix A: Proofs
A.1.1 Proof of Theorem 1
—Proof of Theorem 1 (i)
Let us assume that all the random variables X,X1,…,Xn are defined on the probability space \(({\Omega },\ {\mathscr{A}},\ \mathrm {P}).\) For ω ∈Ω, set
By Theorem 2 of Chacón and Rodríguez-Casal (2010) the set
is of probability 1. For ω ∈Ω, set
and
By the strong law of large numbers P(Ω2) = 1. Let ω ∈Ω1 ∩Ω2. By Lemma 8.3 of Bickel and Freedman (1981), to prove Theorem 1 (i), it is enough to show that
(a) is true because ω ∈Ω1. Let us prove (b). Using the substitution u = (x − Xi)/h we get
We have
Setting
we have
it follows that
From this last inequality we see that \({\Gamma }_{1} \xrightarrow {\quad \quad } 0.\) Hence, (b) is true because ω ∈Ω2.
—Proof of Theorem 1 (ii)
We are going to use the following lemmas in the proof.
Lemma 1.
If (A.2) is true then
Lemma 2.
Let Y be a random variable with cumulative distribution function G. If G is continuous, then \(\mathrm {E}| Y|=+\infty \) if and only if
(Let us continue with the proof of Theorem 1(ii)) It follows from Lemma 1 and Lemma 2 that
Since \(\mathrm {E}| X|=+\infty ,\) by Lemma 2 we have
We have:
A.1.1.1 Proof of Lemma 1
We have
It follows that
This inequality permits to get
A.1.1.2 Proof of Lemma 2
It is well-known that
A.2.1 Proof of Proposition 1
First we prove some useful equalities. Let (xn) be a bounded sequence. Then we have:
Proof of Eq. A.3. We have
Set \(u_{0}=\sup _{n} x_{n}.\) For \(\varepsilon =\frac {1}{k}\ (k\in {\mathbb {N}}^{*}),\ \exists n_{k}\in {\mathbb {N}}\) such that
This last inequality completes the proof of Eq. A.3. The proof of Eq. A.4 is similar. Let α0 ∈ (0, 1) and assume that Q(α0) is the unique solution of the equation F(y) = α0. To prove Proposition 1, it is enough to show that, if (sn) is a sequence then
Proof of the implication (A.5)
–(sn) is necessarily bounded. Suppose for instance that \(s_{n_{k}}\xrightarrow {\quad \quad } -\infty ,\) then \(F(s_{n_{k}})\xrightarrow {\quad \quad } 0 \neq \alpha _{0}.\) This contradicts the fact that \(F(s_{n_{k}})\xrightarrow {\quad \quad } \alpha _{0}.\) Next, set
Using Eq. A.4 we have
Using Eq. A.3 we can prove similarly that
The uniqueness of Q(α0) implies that
and these equalities show that
A.3.1 Proof of Theorem 2
Theorem 2 follows from certain properties of cumulative distribution functions. Lemma 3 below is its restatement in terms of general cumulative distribution functions.
Lemma 3.
Let G and H be continuous cumulative distribution functions. Then we have:
Note that (ii-b) is a restatement of Corollary 1. All the identities in Lemma 3 are presented in Schmid (1993). However, the author states and proves his results assuming that G and H are continuous and bijective. We are going to use the following lemma in our proof.
Lemma 4.
(Lemma 1 in Youndjé (2018)) Let G, H and L be three continuous cumulative distribution functions. Then we have:
Proof of Lemma 3
By Lemma 4 we have:
therefore (i) is established. Since H is supposed continuous, using the substitution α = H(x) (see Apostol (1974)) we have:
These equalities show that (ii-a) is true. Using the substitution α = G(x) we have:
This last equality follows from Eqs. A.6 and A.7. The proof of (ii-b) is obtained from this chain of equalities. Let us set L = (G + H)/2. On one hand we have
by Lemma 3 (ii-b). On the other hand, using Lemma 4, we have
It follows from the two chains of equalities above that
and this identity is exactly Lemma 3 (iii).
A.4.1 Proof of Proposition 2
—Proof of Proposition 2 (i)
We give the proof for the Cauchy cumulative distribution function. The one for the Pareto cdf follows the same steps. We have
thus, to prove Proposition 2 it is enough to show that
Proof of (ii): To prove (ii) it is enough to show that
We have
because of Lemma 1 and Lemma 2. Hence (b) is true. To prove (a), it is enough to show that if F is the Cauchy cdf, we have:
We have
For x > 0, it is widely known that
It follows that
hence, Eq. A.9 is obtained via Taylor expansion of order 1. Using the fact that
the proof of (i) follows the same steps as that of (ii).
—Proof of Proposition 2 (ii) We have:
Thus the proof of Proposition 2 (ii) will be complete if we prove that:
Next, set \(J=[1,\ +\infty )\) and
It is well-known that L2(J) is a vector space, hence to prove (A.10) it is enough to show that
Equation (A.8) shows that 1 − Fh ∈ L2(J). When \(F(x)=1 -\frac {1}{ax^{a}}\) for x > 1 (the Pareto cdf) we have:
and this ends the proof of Proposition 2 (ii).
Rights and permissions
About this article
Cite this article
Youndjé, É. L1 Properties of the Nadaraya Quantile Estimator. Sankhya A 84, 867–884 (2022). https://doi.org/10.1007/s13171-020-00225-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13171-020-00225-0