Skip to main content

Table 2 Summary of three external indices and eight internal indices

From: Clustering Deviation Index (CDI): a robust and accurate internal measure for evaluating scRNA-seq data clustering

 

Index name (abbreviation)

Method

Optimal value

Reference

External

Adjusted Rand Index (ARI)

Rand index adjusted for chance

Max (\(=1\))

[11]

indices

Fowlkes-Mallows index (FM)

Sensitivity and precision

Max (\(=1\))

[12]

 

Normalized Mutual Information (NMI)

Information theory

Max (\(=1\))

[13]

Internal

Calinski-Harabasz index (CH)

Within/between-cluster variance

Max

[14]

indices

Connectivity index (Connectivity)

Clustering consensus of nearest neighbors

Min

[15]

 

Davies-Bouldin index (DB)

Within/between-cluster distance

Min

[16]

 

Dunn index (Dunn)

Within/between-cluster distance

Max

[17]

 

Baker-Hubert Gamma index (Gamma)

Within/between-cluster distance

Max

[18]

 

SD index with average scattering (SD-Scat)

Within/between-cluster variance

Min

[19]

 

Silhouette index (Silhouette)

Within/between-cluster distance

Max

[41]

 

Xie-Beni index (XB)

Within/between-cluster distance

Min

[20]