Table 4.

The estimates of the scaling exponentβfor the power-law model of the size of the clusters The p-value of the corresponding Kolmogorov-Smirnov goodness-of-fit statistic to test the hypothesis that the power-law model is feasible for the size of the clusters generated by all three clustering procedures with α = 0.70. If the p-value is greater than 0.1, we can infer that it is viable that the size of the clusters follows a power-law distribution.

Subject	MCL	IHC	IPC	GMM
	$\hat{\beta }$ (p-value)	$\hat{\beta }$ (p-value)	$\hat{\beta }$ (p-value)	$\hat{\beta }$ (p-value)
1	1.64 (0.273)	1.61 (0.980)	1.58 (0.985)	1.73 (0.591)
5	1.76 (0.603)	1.66 (0.199)	1.59 (0.797)	2.12 (0.882)
6	1.50 (0.003)	1.50 (0.513)	1.50 (0.564)	3.20 (0.977)
7	1.50 (0.216)	1.53 (0.827)	1.50 (0.748)	1.90 (0.774)
8	1.64 (0.713)	1.59 (0.625)	1.51 (0.807)	1.76 (0.938)
10	1.65 (0.041)	1.58 (0.485)	1.56 (0.653)	1.79 (0.981)
12	1.61 (0.642)	1.54 (0.219)	1.56 (0.579)	2.06 (0.727)
13	1.78 (0.423)	1.65 (0.724)	1.53 (0.322)	1.86 (0.781)
15	1.79 (0.609)	1.60 (0.859)	1.61 (0.330)	1.85 (0.256)