CER: Classification Error Rate (CER)

In RSKC: Robust Sparse K-Means

Description

Compute the classification error rate of two partitions.

Usage

CER(ind, true.ind,nob=length(ind))

Arguments

ind	Vector, containing the cluster labels of each case of a partition 1.
true.ind	Vector, containing the cluster labels of each case of a partition 2.
nob	The number of cases (the length of the vector ind and true ind)

Value

Return a CER value. CER = 0 means perfect agreement between two partitions and CER = 1 means complete disagreement of two partitions. Note: 0 <= CER <= 1

Note

This function uses comb, which generates all combinations of the elements in the vector ind. For this reason, the function CER is not suitable for vector in a large dimension.

Author(s)

Yumi Kondo <[e-mail]>

References

H. Chipman and R. Tibshirani. Hybrid hierarchical clustering with applications to microarray data. Biostatistics, 7(2):286-301, 2005.

Examples

vec1<-c(1,1,1,2,3,3,3,2,2)
vec2<-c(3,3,3,1,1,2,2,1,1)
CER(vec1,vec2)

RSKC documentation built on May 2, 2019, 7:23 a.m.