Precomputed forecasts from every successfully fitted insight.cast candidate.
For each state and forecast round, the candidate selected by real-time
cross-validation is duplicated with model_id = "insight.cast".
Format
A data frame with 52,500 rows and 9 columns:
- model_id
Model identifier.
- reference_date
Date defining the forecast round.
- target
Forecast target.
- horizon
Forecast horizon in weeks, from 0 through 3.
- location
Two-digit US location code.
- target_end_date
End date of the predicted week.
- output_type
Forecast output type; always
"quantile".- output_type_id
Quantile probability.
- value
Forecast value.