Skip to main content
. 2016 Nov 17;21(11):1554. doi: 10.3390/molecules21111554

Figure 2.

Figure 2

Comparison of the three methods in handling the data imbalance. The red circle denotes model results based on data sets with the proportion of 1:10 (ligands:decoys). The purple triangle denotes model results based on data sets with the original ratio (1:39). The blue diamond denotes results of models with different weight for ligands class and decoys class in the SVM model training.