Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: xclust
See also: dentoreg cartsplit cartsplitopt cartcv leafnum maketr pred prederr prune prunecv pruneseq prunetot ssr volumes

Quantlet: denvalues
Description: given a binary tree produced by cartsplit, normalizes the mean values of the leaves so that the function represented by the binary tree integrates to one.

Link:
Usage: y = denvalues (tr, dendat)
Input:
tr list of vectors, data structure which represents a binary tree, produced by cartsplit. It contains the vectors tr.val, tr.vec, tr.mean, tr.ssr, tr.nelem, tr.endpoint. See cartsplit for detailed description.
dendat n x p matrix, data matrix of i.i.d. observations. Data with which the density tree was estimated.
Output:
y k x 1 vector, it has as many elements k as there are elements in the vectors forming the tree tr. In the positions which correspond to the leaf nodes there are normalised values of vector tr.mean. Values are normalised in such a way that the function represented by the binary tree integrates to one. In nonterminal positions, the values are +Inf.

Note:

Example:
; loads the library xclust
library("xclust")
; generates the density data
randomize(1)
dendat=uniform(10)~(uniform(10)+1)
; transforms the density data to a regression data
regdat=dentoreg(dendat,4)
; estimates the density tree
tr=cartsplit(regdat.ind,regdat.dep,#(1,1))
; normalises the density estimate
cval=denvalues(tr,dendat)
cval
tr.mean=cval

Result:
Contents of cval
 [ 1,]     +INF
[ 2,]     +INF
[ 3,]     +INF
[ 4,]  0.18356
[ 5,]  0.35149
[ 6,]     +INF
[ 7,]  0.46195
[ 8,]  0.55068
[ 9,]     +INF
[10,]     +INF
[11,]     +INF
[12,]  0.18356
[13,]  0.18356
[14,]     +INF
[15,]  0.18356
[16,]  0.35149
[17,]     +INF
[18,]  0.46195
[19,]     +INF
[20,]  0.18356
[21,]  0.35149



Author: J. Klemelae, 20010912 license MD*Tech
(C) MD*TECH Method and Data Technologies, 05.02.2006