Figure 4From: A Dirichlet process model for classifying and forecasting epidemic curves Accuracy of predicting the epidemic curves in the test set. The accuracy of predicting the epidemic curves in the test set: (a) catastrophic, (b) severe, (c) strong, (d) moderate, (e) mild and (f) milder. Accuracy on day j is the number of curves correctly identified based on the partial epidemic curve on day j. The results are presented based on epidemic peak time. The DP (Dirichlet Process) model is compared to RF (random forest).Back to article page