Table 5: The results of conventional K-means clustering on the data set.

Cluster index	1	2	3	4
Number of samples	1879	3184	174647	20290
Percentage of samples	0.94%	1.59%	87.32%	10.15%