The k-nearest neighbour kernel density estimation method is a special type of thekernel density estimation method with local choice of the bandwidth. An advantage ofthis estimator is that smoothing varies according to the number of observations in aparticular region. The crucial problem is how to estimate the value of the parameterk. In the paper we discuss the problem of choosing the parameter k in a way thatminimizes the value of the asymptotic mean integrated square error (AMISE). Wedene the class of the modied cosine densities that meet the requirements given bythe AMISE. The results are compared in a simulation study.
@article{138,
title = {K-narest neighbour kernel density estimation, the choice of optimal k},
journal = {Tatra Mountains Mathematical Publications},
volume = {51},
year = {2012},
doi = {10.2478/tatra.v50i3.138},
language = {EN},
url = {http://dml.mathdoc.fr/item/138}
}
Orava, Jan. K-narest neighbour kernel density estimation, the choice of optimal k. Tatra Mountains Mathematical Publications, Tome 51 (2012) . doi : 10.2478/tatra.v50i3.138. http://gdmltest.u-ga.fr/item/138/