Skip to main content

Table 5 Prediction results of the random forest classifier using different combinations of features

From: Characterizing key agents in the cryptocurrency economy through blockchain transaction analysis

Groups of features Accuracy Macro Precision Macro Recall Macro F1
Volume 0.488 0.675 0.552 0.469
Temporal 0.804 0.767 0.728 0.730
Network size 0.813 0.752 0.737 0.737
Network structure 0.665 0.610 0.612 0.605
Temporal + Volume 0.844 0.827 0.813 0.814
Size + Volume 0.851 0.820 0.820 0.811
Size + Temporal 0.857 0.854 0.820 0.827
Size + Structure 0.848 0.816 0.787 0.789
Structure + Volume 0.790 0.759 0.752 0.749
Structure + Temporal 0.834 0.839 0.765 0.775
Size + Temporal + Volume 0.867 0.853 0.848 0.845
Structure + Temporal + Volume 0.872 0.867 0.833 0.841
Size + Structure + Volume 0.861 0.828 0.813 0.814
Size + Structure + Temporal 0.882 0.889 0.847 0.850
All groups of features 0.893 0.888 0.862 0.865