Windy Community
    • Unread
    • Categories
    • Groups
    • Go to windy.com
    • Register
    • Login

    Real vs. predicted values

    Scheduled Pinned Locked Moved Your Feedback and Suggestions
    3 Posts 2 Posters 124 Views 1 Watching
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • Jan Suchomel 3J Offline
      Jan Suchomel 3 | Premium
      last edited by

      Hello, I was checking the premium feature of comparing models predictions vs. the actual outcome but I noticed that some of the values I see in those comparisons I have never seen before as a prediction from those models... For example in the comparison it is claimed that Meteoblue predicted 8C around 12 in NYC local time but I am quite sure I have not seen this value predicted before. Since the models are always recalculating the predictions when is the cutoff for the predictions to be taken for the comparison?

      SutyS 1 Reply Last reply Reply Quote 1
      • SutyS Online
        Suty Windy Staff @Jan Suchomel 3
        last edited by

        @Jan-Suchomel-3 Hello, to be honest and fair with you, this is difficult to tell. I would rather suggest comparing all available models to get the most relevant prediction.

        1 Reply Last reply Reply Quote 0
        • Jan Suchomel 3J Offline
          Jan Suchomel 3 | Premium
          last edited by

          Hmmm considering its one of the premium features it is quite disappointing... Especially since it feels like the models with more frequent recalculations score better which would suggest the last forecast/recalculation before observation is taken into account...

          If anything would be possible I would split it to several "time windows" and you could choose the cut off period you like and compare how precise the models were 7,3 and 1 day before or even 12hours or something like that...

          I understand it might be to complex but at least the cutoff period should be unified for all models to the same time so the comparison is not as skewed as it is now....

          1 Reply Last reply Reply Quote 0
          • First post
            Last post
          Windy Community  |  Powered by excellent NodeBB
          Terms of Use     Privacy Policy     Windy.com