Table 6 Distribution of the differences \(|k_{true}-k_{opt}|\) across all measures values, separately for real and artificial data.

From: Ground truth clustering is not the optimum clustering

Measure

Real data

Artificial data

\(|k_{true}- k_{opt}|\)

\(|k_{true}- k_{opt}|\)

0

1

2

0

1

2

(%)

(%)

(%)

(%)

(%)

(%)

AMI

25.00

8.33

66.67

25.00

16.67

58.33

ARS

25.00

16.67

58.33

58.33

8.33

33.33

h

16.67

25.00

58.33

16.67

0.00

83.33

c

8.33

8.33

83.33

33.33

33.33

33.33

NMI

8.33

8.33

66.67

41.67

0.00

58.33

FMS

50.00

25.00

25.00

58.33

8.33

33.33

CHC

41.67

16.67

41.67

25.00

16.67

58.33

DBI

25.00

41.67

33.33

33.33

41.67

25.00

\(S_{score}\)

25.00

33.33

41.67

41.67

25.00

33.33