Table 9.

Detailed performance comparison of random forest classifiers based on three sets of features, built on the base data set. Metrics are calculated on the test data. All results are rounded to three decimal digits. For each classifier, the test set consists of 12,000 trajectories per diffusion mode—that is, 36,000 in total.

Set of Features	Measure	Normal Diffusion	Subdiffusion	Superdiffusion	Total/Average
	Precision	0.929	0.973	0.970	0.957
Set A	Recall	0.944	0.966	0.962	0.957
	F1	0.936	0.969	0.966	0.957
	Precision	0.910	0.970	0.963	0.948
Set B	Recall	0.934	0.957	0.950	0.947
	F1	0.922	0.964	0.956	0.947
	Precision	0.912	0.969	0.966	0.949
Set C	Recall	0.935	0.958	0.951	0.948
	F1	0.923	0.963	0.959	0.948