Confidence Interval

To toggle the Prediction's Confidence Interval bounds of the in the chart, click in the Prediction's legend or line

Interval Bounds

When the Confidence Interval is active, the Prediction will display the 50% and 90% bounds (upper and lower)

MAE

Measures the average magnitude of the errors in the predictions. Lower values mean predictions are closer on average to actual data

MSE

Measures the average of the squared differences between forecast and observation. Lower values mean fewer and/or smaller large errors in predictions

CRPS

Measures how well the predicted cumulative distribution function (CDF) matches the observed outcome. Lower values indicate the forecast distribution better matches the observed outcome

Log Score

Measures the log-probability assigned to the observed value by the forecast distribution. Higher log-probabilities mean the forecast assigned more probability to the actual observation

Interval Score

Evaluates prediction intervals (e.g., 90% confidence intervals) of the forecast. Lower values mean narrower prediction intervals that still capture the true value appropriately