Skip to main content
. 2020 Jun 15;8(3):215–250. doi: 10.5599/admet.835

Table 5.

Statistical evaluations of predictions for the Extra Trees, Random Forest and Bagging predictors and for the Vox Machinarum consensus classifier over Avdeef’s average log S values [45] for all 32 compounds comprising the 2019 Solubility Challenge loose test set of 32 molecules. The Vox Machinarum predictions reported here were the median of the other three classifiers’ predictions for each compound. The standard deviation of the 32 compounds’ log S values was 2.142.

Method RMSE RMSE/SD AAE R 2 Err < 0.5 Err < 1.0
Extra Trees 1.517 0.708 1.103 0.680 10 (31%) 19 (59%)
Random Forest 1.495 0.698 1.109 0.700 10 (31%) 18 (56%)
Bagging 1.549 0.723 1.160 0.708 8 (25%) 17 (53%)
Vox Machinarum 1.490 0.696 1.097 0.712 11 (38%) 18 (56%)