You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A common question is regarding how to interpret the results in “plain language”. That’s maybe not necessary for the more standard metrics (MAE, MBE, RMSE), but it does seem beneficial to make that clearer to report readers. One option is revising the metrics page (as needed) so each metric has an interpretation statement and then linking to the metrics page in the report. Another option would be adding a hover box or maybe just bullet points that list the interpretations per metric. E.g., MBE measures the bias of a forecast; MBE < 0 means the forecast tends to under-predict, MBE > 0 means the forecast tends to over-predict, and MBE = 0 means the forecast has no bias overall.
We should definitely add links from the report back to the metrics page. I like the idea of an interpretation statement, but worry that it would take a lot of time to do right. Hover boxes might also be helpful if we can keep them short and sweet.
Same applies to summary statistics.
Part of my interest in proposing this project came from struggling with the RMSE decomposition for solar forecasts. I've always wanted the reports to explore those kinds of issues. One challenge is that the metrics are user-selected and not always present in the report. I'll have to think harder about how to handle that. There has to be a better answer than just put it on the static metrics page - otherwise that'll turn into a text book that no one reads.
The text was updated successfully, but these errors were encountered:
From @dplarson and @awig
We should definitely add links from the report back to the metrics page. I like the idea of an interpretation statement, but worry that it would take a lot of time to do right. Hover boxes might also be helpful if we can keep them short and sweet.
Same applies to summary statistics.
Part of my interest in proposing this project came from struggling with the RMSE decomposition for solar forecasts. I've always wanted the reports to explore those kinds of issues. One challenge is that the metrics are user-selected and not always present in the report. I'll have to think harder about how to handle that. There has to be a better answer than just put it on the static metrics page - otherwise that'll turn into a text book that no one reads.
The text was updated successfully, but these errors were encountered: