This directory gives the key SVC performance data - i.e the calculated balanced accuracies when predicting galunisertib h bond classifications or ROY angle classifications from ml descriptors.

SV_5 in a filename denotes that the data is for the predictions made using the best performing (gives the best balanced accuracy) combination of up to the top five ranked ML descriptors

SV_single in a filename denotes that the data is for the predictions made using the best performing (gives the best balanced accuracy) single descriptor the top 32 ranked ML descriptors

c_x in a filename denotes the value of paramater c used in the model (p1 is 0.1)


data is presented as arrays, each element is the balanced accuracy of predictions for different SOAP cut-offs : 4,6,8, then 10 angstroms.

