Fig. 4.
(A) MLR model, using the same descriptors as the Fig. 3 model, used to predict rates of external validation amide couplings. (B) Model retrained on all training and external validation data (pseudorandom 70:30 training/test split), where x indicates a coupling previously in the external validation set. (C) External validation couplings with their measured rates and predicted rates based on the MLR model in A and B. (D) Living model concept schematic.