Predicting Ovarian/Breast Cancer Pathogenic Risks of Human BRCA1 Gene Variants of Unknown Significance

Figure 2

The overall performance of optimized support vector machine, optimized random forest model, original (not optimized) support vector machine, and original random forest. The optimized random forest (dark blue) had an obviously larger AUC than the optimized support vector machine (light blue) and the original random forest (purple), while no significant increase of AUC was observed between the original support vector machine (yellow) and the optimized one (light blue). The quantified AUC values of optimized random forest and optimized support vector are 0.85 and 0.75, respectively, indicating that the random forest model had better performance after optimization, while the support vector machine did not.