Validation and Evaluation

Note

This page is under development.

• What to check first: visual inspection of past nowcasts against final observed values (flipbooking, overlay plots)
• What “good enough” looks like: prediction intervals cover observed values at roughly the expected rate (e.g. 90% intervals contain ~90% of final values)
• When to investigate further: persistent directional bias, intervals that are consistently too narrow or too wide, sudden changes in performance
• Applied quantitative methods (coverage, correlation, residuals)
• Advanced methods (WIS, MAE/MSE) for comparison
• Evaluation for public health utility
• Against a baseline and other common methods