MultivariateRandomForest (version 1.1)

spliting: Split of the Parent node

Description

Split of the training samples of the parent node into the child node for the feature which gives the minimum cost of splitting

Usage

spliting(X, Y, mtree, Index, V_inv, Command)

Arguments

X

Input Training matrix of M x N, M is the number of training samples and N is the number of features

Y

Output Training response of M x T, M is the number of samples and T is number of ouput Features(Response)

mtree

number of randomly selected features used for each split

Index

Index of training samples

V_inv

Covariance matrix of Output Feature matrix

Command

1 for RF and 2 for MRF depending on the method

Value

List with the following components:

index_left

Index of the samples which are in the left node after splitting

index_right

Index of the samples which are in the right node after splitting

which_feature

The number of the feature which gives the minimum splitting cost

threshold_feature

The threshold value, which will decide a sample will go to the left node or, right node. If the training sample value is less than threshold value, it will go to the left node or if greater then go to the right node.

Details

In a node of a decision tree, there are number of samples with different features. In time of splitting, a fixed number of features(mtree) has been selected randomly(that's why it is called random forest). For the splitting, node cost for all the spliting of these features are considered and whichever gives the minimum value has been selected as the splitting criteria(feature value and threshold value of the feature) of this node split.

Examples

X=matrix(runif(20*100),20,100)
Y=matrix(runif(20*3),20,3)
mtree=5
Index=1:20
V_inv=stats::cov(Y)
Command=2#MRF, as number of output feature is greater than 1
Split_criteria=spliting(X,Y,mtree,Index,V_inv,Command)