Skip to main content

Table 6 Performance measures for binary classification of mixin models, Q refers to relative quantity of NA compounds added to the training data

From: Nonadditivity in public and inhouse data: implications for drug design

ChEMBL data

RF (MCC for test)

Q0 (0.0%)*

Q1 (0.6%)*

Median (1.3%)*

Q3 (2.6%)*

1613777

DTC-split

0.02

0.04

− 0.04

0.02

 

All-split

0.00

0.04

− 0.05

− 0.02

1613797

DTC-split

− 0.03

0.07

0.20

− 0.12

 

All-split

0.00

0.00

0.00

0.00

1614027

DTC-split

0.28

0.28

0.28

0.20

 

All-split

0.22

0.04

− 0.05

0.11

  1. * Test set size for Q0 differs from Q1/Median/Q3.