Skip to main content
. 2019 May 24;6:36. doi: 10.3389/fmolb.2019.00036

Figure 10.

Figure 10

Coefficients of the LR model in the test set averaged over 200 different splitting of the training and external test sets for the A and C data-sets (A) and for the LR built on molecular descriptors only (B), as denoted in each plot.