Repository navigation
Dependencies
Refreshed renv.lock to pull the latest releases of the hubverse packages the
image wraps, hubPredEvalsData 1.2.0 -> 1.3.0 and hubEvals 0.3.1 -> 0.5.0,
together with scoringutils 2.2.0 -> 2.3.0 and the other CRAN dependencies as
of the 2026-10-05 CRAN snapshot (#51).
The user-visible changes these bring to the generated dashboard data:
- Relative-skill scoring is substantially faster and uses much less memory,
which matters most on hubs with many models, such as FluSight. Relative
skill values are unchanged (hubverse-org/hubEvals#144,
epiforecasts/scoringutils#1221). - The number of predictions scored (
n) inscores.csvnow counts only
forecasts that had a corresponding observation to score against, rather
than every submitted prediction. This corrects an overcount wherever a
model submitted predictions for units with no observation. In the rare
case where this count differs across a target's output types,scores.csv
reports a separaten_<output_type>column per output type in place of
the singlen(hubverse-org/hubPredEvalsData#19). - Hubs whose target is identified by
target_idalone, with no target key
task id variable (target_keys: null), as used by nowcast hubs, are now
supported. Previously these failed partway through the run
(hubverse-org/hubPredEvalsData#87). - Evaluation data generation is faster on every hub: the hub connection is
opened once and reused across targets and evaluation sets
(hubverse-org/hubPredEvalsData#82).
Base and dev images
- Refreshing
renv.lockwithscripts/update.Rnow moves the CRAN
dependencies to their current releases together with the hubverse
packages. Previously CRAN packages stayed at the versions the rocker base
image was built against, so a hubverse release that required a newer CRAN
package could not be picked up (#51).