Journal of Mathematical Sciences: Advances and Applications
Author's: Hamza Dhaker, Papa Ngom, El Hadji Deme and Malick Mbodj
Pages: [57] - [83]
Received Date: April 25, 2018; Revised May 14, 2018
Submitted by:
DOI: http://dx.doi.org/10.18642/jmsaa_7100121962
The choice of bandwidth is crucial to the kernel density estimation
(KDE). Various bandwidth selection methods for KDE, least squares
cross-validation (LSCV) and Kullback-Leibler cross-validation are
proposed. We propose a method to select the optimal bandwidth for the
KDE. The idea behind this method is to generalize the LSCV method,
using the measure of 
and to see the improvement in our method, we
will compare these 
bandwidth selector with a normal reference
(NR), the last squares cross-validation (LSCV), the Sheather and Jones
(SJ) method, and the generalized 
bandwidth selector, on simulated data. The use
of the various practical bandwidth selectors is illustrated on a real
data example.
nonparametric density estimation, measure divergence, integrated squared error, bandwidth.
