Figure 2From: Classification of rheumatoid arthritis status with candidate gene and genome-wide single-nucleotide polymorphisms using random forestsROC curve of five-fold CV using RFs with the 500 most important SNPs. For each CV, a prediction model is built by using the training dataset and the ROC curve is generated by comparing the predicted RA status with the true RA status in the testing dataset. Each color curve represents prediction accuracy of one of the five CVs.Back to article page