Score summaries comparing insight.cast with official FluSight team submissions.
They let the vignette run without refitting models or downloading Hub data.
Format
flusight_scores is a data frame with one row per model and these
columns:
- model_id
Model identifier.
- wis
Mean weighted interval score.
- interval_coverage_50
Empirical coverage of the 50 percent interval.
- interval_coverage_95
Empirical coverage of the 95 percent interval.
- wis_relative_skill
Pairwise relative WIS.
- wis_scaled_relative_skill
Relative WIS scaled to the
FluSight-ensemble; values below one are better.- count
Number of forecast tasks scored.
- model_group
One of
"FluSight ensemble","FluSight baseline","FluSight model", or"insight.cast model".- display
Whether the vignette's concise comparison displays the model. All models remain available for custom filtering.
Source
Forecasts and target data from the FluSight Forecast Hub, version 1.2.0.
Details
flusight_scores contains scores pooled across locations.
flusight_scores_by_location contains scores stratified by location.