Cluster_Index

Several clustering indexes implementation

External Indices

This indexes are designed to measure the similarity of two partitions.

Indexes based on pair counting

All this indices are based on counting pair depending on wether they belong to the same cluster or not according to the partition C or C'. There are four possibilities:

• yy : the two points belong to the same cluster according to both C and C'
• yn : the two points belong to the same cluster according to C but not to C'
• ny : the two points belong to the same cluster according to C' but not to C
• nn : the two points does not belong to the same cluster according to C and C'

Precision

Recall

Rand Index [3]

Adjusted Rand Index

Folkes-Mallows Index [4]

Jaccard Index

Kulczynski Index

McNemar Index

Phi Index

Rogers and Tanimoto Index

Russel and Rao Index

Solkal and Sneath Index (version 1 and 2)

Hubert Index

Mirkin Metric

Measures based on set overlaps

F-measure

Also call Czebanowski-Dice index or Ochiai index

Measures based on Mutual Information

Mutual Information

Strehl and Ghosh Normalized Mutual Information

Fred and Jain Normalized Mutual Information

Variation of Infomation

Other

Purity

Entropy

References

1. Silke Wagner and Dorothea Wagner, Comparing Clustering - An Overview, 2007
2. Bernard Desgraupes, Clustering Indices, 2016, [https://CRAN.R-project.org/package=clusterCrit]
3. Rand, William M.: Objective Criteria for the Evaluation of Clustering Methods. Journal of the American Statistical Association, 66(336):846-850, 1971.
4. Fowlkes, E. B., Mallows, C. L.: A Method for Comparing Two Hierarchical Clusterings. Journal of the American Statistical Association, 78(383):553–569, 1983.

TODO

• add internal cluster indexes
• complete References section
• add definition and formula for each indexes