Estimation de densité (density estimation):

en lien avec l'article :
- Optimal Bandwidth location for local bandwidth kernel based density estimation. C. Aaron, soumis en janvier 2009 a Computational Statistics & Data Analysis preprint : ps

Download the 3 following files
    bandwidth_SJ  : compute the bandwidth according to the Sheater and Jones Algorithm (not my program picked up on     ttp://www.mathworks.com/matlabcentral/fileexchange/)
    LCVgauss : compute the cross-validate ise
    estimcatSJgr6 : density estimation function :
    function [dens0,dens,plage,C,S]=estimcatSJgr6(X,nbplage)

    inputs :
    X : is the data set on which the density is estimated (that should be sorted) containing N poins
    nbplage : is the number of point on which the density is computed

    outputs :
    dens0 : density estimation for a global bandwidth (SJ method)
    dens : density estimation for at most p local bandwidths (p is entered by the user when asked : it should be the number of local minima of         dens0)
    plage : is the nbplage points where the density is estimated
    S : is the bandwidths values
    C : is the associated clustering (C(i)=k if the bandwidth associated to X_i is S_k)


 retour à l'accueil