grow {densup}R Documentation

grows a large multitree whose nodes represent certain bins of the sample space

Description

Given density data creates a large multitree whose terminal nodes represent certain bins of a rectangle containing the data. Nodes are annotated with frequencies. The rectangle containing the data is partitioned to bins which are represented with the terminal nodes of the tree.

Usage

grow(dendat, N, h)

Arguments

dendat n*d data matrix
N vector of length d; for each direction number of bins in the finest partition; should dyadic numbers 8,16,32,64,...
h nonnegative number; determines how much the smallest rectangle containing the data will be blown up to get the rectangle which will be partitioned to bins

Value

list containing vectors

child link to the first child
sibling ling to the next sibling
parent link to the parent
location gives for terminal nodes the number of bin in question
direction gives the direction of split for the nodes with 2 children, for the nodes with d children equals to 0
frekv number of observations in the bin corresponding to the node
volume volume of the bin corresponding to the node
floor pointers to the terminal nodes of the tree
upp upper bin number contained in the bin corresponding to the node
low lower bin number contained in the bin corresponding to the node; for terminal nodes upp=low

Author(s)

Jussi Klemelä

References

http://www.rni.helsinki.fi/~jsk/densup

See Also

plotsupport

Examples

dendat<-matrix(rnorm(20),10)
N<-c(8,8)
h<-0.1
tree<-grow(dendat,N,h)
[Package Contents]