Table 4 Comparison between DBSCAN and KM-DBSCAN across multiple parameter settings.

From: KM-DBSCAN: an enhanced density and centroid based border detection framework for data reduction towards green AI

Method

\(\varepsilon\)

MinPts

ARI

Time (s)

 

2.75

399

0.91

0.0249

 

2.76

399

0.00

0.0240

 

2.75

405

0.913

0.0229

DBSCAN (whole data set = 1500)

2.77

405

0

0.0269

 

2.77

406

0

0.0229

 

2.80

407

0

0.0249

 

2.80

415

0.902

0.0229

 

1.97

2

0.979

0.001

 

1.97

3

0.979

0.001

 

1.98

3

0.979

0.001

KM-DBSCAN (k=50)

1.98

2

0.979

0.001

 

1.99

2

0.979

0.001

 

2.00

4

0.979

0.001

 

2.00

5

0.979

0.001

 

2.30

14

0.951

0.001

 

2.31

14

0.955

0.001

 

2.31

15

0.964

0.001

KM-DBSCAN (k=100)

2.33

14

0.967

0.001

 

2.34

14

0.967

0.001

 

2.34

16

0.958

0.001

 

2.35

16

0.958

0.001

 

2.60

36

0.940

0.001

 

2.61

36

0.940

0.001

 

2.62

35

0.945

0.001

KM-DBSCAN (k=150)

2.63

35

0.945

0.001

 

2.64

35

0.953

0.001

 

2.64

36

0.943

0.001

 

2.63

37

0.938

0.001

Back to article page