Zero-Bias Locally Adaptive Density Estimators (Q4455911)
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 2059355
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | Zero-Bias Locally Adaptive Density Estimators |
scientific article; zbMATH DE number 2059355 |
Statements
Zero-Bias Locally Adaptive Density Estimators (English)
0 references
16 March 2004
0 references
kernel PDF estimator
0 references
bandwidth
0 references
cross-validation
0 references
0.8626231551170349
0 references
0.8363105058670044
0 references
0.8363102078437805
0 references
0.810614824295044
0 references
0.8067086338996887
0 references
A kernel PDF estimator \(\widehat f(x)\) is considered with locally adaptive bandwidth \(h(x)\) of the form \[ \widehat f(x)= (nh(x))^{-1} \sum_{i=1}^n K ((x-x_i)/h(x)), \] where \(x_i\) are i.i.d. with PDF \(f\). It is shown that for many \(f\) there exists a bandwidth function \(h\) such that \(E_f\widehat f=f\), i.e., \(\widehat f\) is an unbiased estimator of this fixed \(f\). The authors propose an estimate \(\widehat h\) for such ``unbiased'' bandwidths based on the cross-validation technique. The asymptotic behaviour of \(\widehat h\) is investigated. Simulation results and application to medical data are presented.
0 references