Table 6.

F-scores for the features that yielded the best prediction results are given. Out of 84 graph metrics, connectivity measures such as giant connected component ratio, clustering coefficient, isolated points and normalized Laplacian metrics are the most discriminative features.

Green Giant Connected Ratio	1.43
White Clustering Coefficient	1.11
White Giant Connected Ratio	1.07
White Percentage of Isolated Points	1.07
Red Clustering Coefficient	0.98
Number of 2s in Red Normalized Laplacian	0.93
Blue Clustering Coefficient	0.89
Number of 1s in Red Normalized Laplacian	0.89
Red Normalized Laplacian Energy	0.87
Red Percentage of Isolated Points	0.82
Blue Percentage of Isolated Points	0.81