The Equivalence of Weak, Strong and Complete Convergence in $L_1$ for Kernel Density Estimates
Devroye, Luc
Ann. Statist., Tome 11 (1983) no. 1, p. 896-904 / Harvested from Project Euclid
Let $f$ be a density on $R^d$, and let $f_n$ be the kernel estimate of $f$, $f_n(x) = (nh^d)^{-1} \sum^n_{i=1} K((x - X_i)/h)$ where $h = h_n$ is a sequence of positive numbers, and $K$ is an absolutely integrable function with $\int K(x) dx = 1$. Let $J_n = \int |f_n(x) - f(x)| dx$. We show that when $\lim_nh = 0$ and $\lim_nnh^d = \infty$, then for every $\varepsilon > 0$ there exist constants $r, n_0 > 0$ such that $P(J_n \geq \varepsilon) \leq \exp(-rn), n \geq n_0$. Also, when $J_n \rightarrow 0$ in probability as $n \rightarrow \infty$ and $K$ is a density, then $\lim_nh = 0$ and $\lim_nnh^d = \infty$.
Publié le : 1983-09-14
Classification:  Nonparametric density estimation,  $L_1$ convergence,  kernel estimate,  strong consistency,  60F15,  62G05
@article{1176346255,
     author = {Devroye, Luc},
     title = {The Equivalence of Weak, Strong and Complete Convergence in $L\_1$ for Kernel Density Estimates},
     journal = {Ann. Statist.},
     volume = {11},
     number = {1},
     year = {1983},
     pages = { 896-904},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1176346255}
}
Devroye, Luc. The Equivalence of Weak, Strong and Complete Convergence in $L_1$ for Kernel Density Estimates. Ann. Statist., Tome 11 (1983) no. 1, pp.  896-904. http://gdmltest.u-ga.fr/item/1176346255/