Figure 1From: Evaluation of prediction models for the staging of prostate cancer Illustration of 5-fold cross validation. The dataset is randomly split into five subgroups (called folds). Four folds are merged together to form a training dataset. The 5th fold is used as the testing dataset. This process is then repeated five times.Back to article page