Post by iconPost by DonJuan21 | 2019-07-10 | 00:46:08

Here's a graph of the zezo predicted finish times based on a snapshot of the forecast times. I've included an exponential decline in the past forecast's value in the weighted average.

Does that make sense?

https://i.postimg.cc/Xv1WTyj7/Marseille.jpg

commenticon 5 Comments
Post by iconPost by DonJuan21 | 2019-07-10 | 00:47:16

Post by iconPost by zezo | 2019-07-10 | 19:50:39
I have to add sub-hour resolution next time.

But no matter what, the winning run always seems to start around 4AM my time ;-)
Post by iconPost by DonJuan21 | 2019-07-23 | 21:40:28
If you do add half hour starting times resolution that would be cool.

http://bit.ly/finishtimes

That's my revised finishing times chart that will update when your chart does.

It shows both full pack and no options as well as the current record to beat.

DonJuan21
Post by iconPost by zezo | 2019-07-29 | 19:12:55
There is a little problem with that. The router runs in 1-hour steps by the time it reaches the finish. Taking the residual DTF into account is not exactly an option, because the final point tends to be all over the places, sometimes past the finish mark.

Anyway, the 1-week uncertainty in weather should much more than the 1-hour resolution and it's better not to create false feeling of accuracy by adding precision to the display.
Post by iconPost by DonJuan21 | 2019-08-02 | 06:40:32
It was interesting getting the exponential decay constant right. If it is too gradual, it shows oscillations when the forecast "n" get's slim, like about a 48hrs out.

To make the blend from forecast to forecast not be oscillatory and model realistic non-oscillatory transitions, it can't take much more than 24hrs into account.

That is, the half-life reliability of any particular forecast is about 4-5 hrs. A 24hr old forecast can't be trusted at the 3% level. A 6hr old forecast is worth about 40% confidence.

Wish we could remove those discontinuities.

You can see that when you play the game. About half the time, there's a noticeable shift as forecasts transition.

DJ

More Cow Bell, more resolution.

border
Topics list
Posts
border
1
border
border
5
border
border
Copyright 2009 by ZEZO.ORG. All Rights Reserved.