No label defined (Q4238469)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1271138
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 1271138

      Statements

      0 references
      1 May 2001
      0 references
      asymptotics
      0 references
      efficiency
      0 references
      multiresolution approximation
      0 references
      projection kernels
      0 references
      This paper is devoted to the estimation of a density function \(f\) based on i.i.d. observations \(X_1,\dots,X_n\). The author defines his estimator as \(\widehat f(x)=n^{-1}\sum_{i=1}^{n}K(x,X_i)\), where \(K(x,y)\) is positive definite and satisfies \(K(x,y)=K(y,x)\). He focuses his attention on wavelet expansions. If \(\{V_j\}_{j\in Z}\) is a nested sequence of subspaces of \(L^2(R)\), \(\phi\) and \(\psi\) are the scaling function and the wavelet, respectively, the projection of \(f\in L^2(R)\) onto \(V_j\) is given by \(P_hf(x)=\int_{-\infty}^{\infty}K_h(x,y)f(y)dy\), where \[ K_h(x,y)=h^{-1}K(x/h,y/h)\;\text{with} K(x,y)=\sum_{k\in Z}\phi(x-k)\phi(y-k)\;\text{and} h=2^{-j},\;j\in Z. \] A density estimator based on projection kernels proposed in the literature is \[ \widehat f(x)=(nh)^{-1}\sum_{i=1}^{n}K(x/h,X_i/h). \] Introducing the notion of kernels of order \(m\) and characterizing the wavelet expansions that have symmetric multiresolution approximations, the asymptotic bias is given in Theorem 3.2 and the asymptotic variance in Theorem 4.1. Combining both results, the optimal choice and the evaluation of the M.I.S.E. can be obtained. Two other subjects considered are: bias reduction using an average kernel estimator and an efficiency study comparing different kernels.
      0 references

      Identifiers