Table 9.

Random Forest and Dummy classifiers balanced accuracy of preprocessed GED with discretizations pipelines vs. $lo{g}_2$ GED (accuracies are expressed as percentages)

Pipeline	RF Bal. Acc.	Dummy Bal. Acc.
tSNE Uniform	$61.4\pm 9.4$	$17.6\pm 3.4$
tSNE Log-z	$58.4\pm 9.3$	$17.4\pm 3.9$
tSNE Normal	$62.5\pm 9.0$	$13.9\pm 2.8$
UMAP Uniform	$61.4\pm 9.4$	$16.9\pm 2.8$
UMAP Log-z	$58.5\pm 8.5$	$13.3\pm 3.5$
UMAP Normal	$62.5\pm 9.0$	$14.2\pm 2.3$
$lo{g}_2$ GED	$26.0\pm 3.1$	$17.7\pm 4.5$