search

USGS-R / protoloads

Prototyping and exploring options for broad-scale load forecasting
0 stars 4 forks source link

rerun models and figures #74

Closed aappling-usgs closed 6 years ago

aappling-usgs commented 6 years ago

Reran the models with the correct discharge data and changed gather_forecasts back to not use the temporary hack. This led to changes in the hashes but I hope not in the important column contents of preds_wrtds and preds_loadest...but I'm in vacation mode and not entirely sure I've done the right checks to really confirm.

Jake, since you have the old preds_x.rds files on your computer, could you stash those and compare to the new ones on Drive before merging this? The magnitudes should be correct and basically the same as what we had before.

I also rebuilt the figures, which could serve as a second source of checks. They look right to me, but Jake, could you take a look at those too to see if there's anything funny now?

jzwart commented 6 years ago

Comparing the new model run output to the old model output with the temporary hack looks good for both Loadest and WRTDS runs: image image

The updated figures look good too!

aappling-usgs commented 6 years ago

Phew! Thanks for checking.