Clustering Analysis of Multivariate Data: A Weighted Spatial Ranks-Based Approach

Table 3

Comparison of different clustering approaches applied to the old faithful dataset.

Clustering method	No. of clusters	Cluster sizes	H	Purity	Entropy	ARI
WSR	2	172, 100	0.018	0.98	0.12	0.93
GMM (mclust) “BIC”	3	40, 97, 135	0.15	1	0	0.71
K-means “CH index”	10	46, 42, 19, 23, 48, 21, 23, 13, 20, 17	0.71	0.98	0.07	0.18
HDDC “BIC”	2	175, 97	0	1	0	1
MixtPPCA “BIC”	2	174, 98	0.004	0.996	0.03	0.98
PAM “silhouette width”	2	172, 100	0.018	0.98	0.12	0.93
DBSCAN	3	168, 82, 17 (5 noise points)	0.08	0.98	0.09	0.85
KMD	3	97, 76, 99	0.28	1	0	0.60
FCM	2	172, 100	0.018	0.98	0.12	0.93
GG	2	97, 175	0	1	0	1
DDC	2	172, 100	0.018	0.98	0.12	0.93
SNN	2	270, 2	0.01	0.64	0.94	−0.01
densityClust	1	272	0	0.64	0.94	0

Results are based on the mean of 1000 repetitions.