Add a mean squared error metric to display the difference between predicted and observed values.
-
In the "Observation vs. forecast" tab it would be really cool to have an index (i'm proposing MSE to keep it simple) that shows how close a particular model was to actual data in the last 10 days, for temperature, wind direction, speed and maybe humidity as well. This would make it easy to learn which model works best for that particular area.
If this is already present, or if you tell me you'll consider adding something along these lines, I'm buying premium immediately -
@pol9
The error of each model relative to the reported parameter is already displayed in the browser version of the Observation vs forecast feature.
It is the median error more exactly the Median Absolute Error, different from the Mean Squared Error you mean. It is the median difference between the observations (true values) and model output (forecast).
With this new feature, you easily find which is the best model according to temperature or wind in the area of the weather station.
-
@pol9
Furthermore now in the browser version, you can select the category of weather station you want. This is an great add-on to relying on professional weather stations.
-
@idefix37 That is amazing! I will stick to my promise and buy premium.
When I was on that tab and the idea came to me, for a moment I saw the blurred part on the right and I thought that it might have been what I was looking for. But since I found no mention of this new feature in the two most recent updates (I didn’t look hard enough) I thought it maybe was a recap of the different models. May I suggest unblurring one of the models error tab so it’s more clear what they are? Thanks a lot! -
I idefix37 referenced this topic on
