Fig. 3
From: Leveraging TCGA gene expression data to build predictive models for cancer drug response
Random forest classifier performance for pan-cancer models. a Comparison of the cross-validation predicted probabilities between non-responders and responders for the 5-FU pan-cancer model. b Comparison of the cross-validation predicted probabilities between non-responders and responders for the GCB pan-cancer model. c ROC curve for 5-FU pan-cancer model; Cross-validation (Sensitivity: 0.97 Specificity: 0.66 AUC: 0.98) Model-validation (Sensitivity: 0.64 Specificity: 0.33 AUC: 0.56). d ROC curve for GCB pan-cancer model; Cross-validation (Sensitivity: 0.80 Specificity: 1.0 AUC: 0.98) Model-validation (Sensitivity: 0.70 Specificity: 0.94 AUC: 0.71)