Sparse Optimal Discriminant Clustering

Share:

Description

To perform Sparse Optimal Discriminant Clustering

Usage

1
2
sodc.clust(x, centers, l1 = -1, l2 = -1, cv.num = 5, clus = kmeans, boot.num = 20, 
l2.idx = seq(-3, 3, by = 6/20), l1.idx = seq(-3, 3, by = 6/20))

Arguments

x

A numberic dataset matrix.

centers

An integer scalar with the desired number of groups.

l1

L1 penalty parameter. The larger lambda1, the sparser result. if l1==-1, will automatically select optimal lambda1.

l2

L2 penalty parameter. if l2==-1, will automatically select optimal lambda2.

cv.num

The numbers of cross validation iteration for selecting tuning parameter lambda2.

clus

The clustering method applied on SODC component.

boot.num

The numbers of bootstrap iteration for selecting tuning parameter lambda1.

l2.idx

A sequence of index numbers from which one can get a sequence of lambda2 values by calculating 10^l2.idx.

l1.idx

A sequence of index numbers from which one can get a sequence of lambda1 values by calculating 10^l1.idx.

Value

cl

An object of class "kmeans" or other class depending on the value of argument "clus" using SODC component.

clvar

An object of class "kmeans" or other class depending on the value of argument "clus" using SODC selected varialbes.

opt.lambda1

optimal lambda1 selected. If l1!=-1, will return opt.lambda1=l1.

opt.lambda2

optimal lambda2 selected. If l2!=-1, will return opt.lambda2=l2.