| Library: | smoother |
| See also: | binlindata Rdenbest denbbwcrit |
| Quantlet: | binweights | |
| Description: | direct computation of the autocovariances of the bincounts needed for fast computation of the kernel estimates of the integrated squared density derivatives. |
| Usage: | binw = binweights(binc) | |
| Input: | ||
| binc | m x 1 vector, corresponding to the bin counts as output from binlindata or a similar procedure | |
| Output: | ||
| binw | m x 1 vector, autocovariances of bin counts | |
library("smoother")
library("xplore")
n = 1000
d = 0.1
{w,mu,sigma}=normalmixselect("Marron_Wand_8")
x = normalmix(n,w,mu,sigma)
; x = round(x,1)
{bing,binc}=binlindata(x,d)
binw = binweights(binc)
bing~binc~binw
n~sum(binc)
(binw[1]+2*sum(binw[2:rows(binw)])) /(n*n)
Generates 1000 variates from a normal mixture example density, optionally rounds them, computes a bin grid, the corresponding bincounts and the bin weights. The bin counts sum up to the total sample size and, as a check, the last expression should be 1.