Smart Energy Supply Scheduling for Green Telecom Challenge by ITU
18 000 CHF
Completed (over 1 year ago)
Prediction
833 joined
158 active
Start
Aug 06, 24
Close
Sep 30, 24
Reveal
Oct 03, 24
'%3e%3cpath fill='%23001489' d='M0 0v6h9V0z'/%3e%3cpath fill='%23e03c31' d='M0 0v3h9V0z'/%3e%3cg stroke='%23fff' stroke-width='2'%3e%3cpath id='d' d='m0 0 4.5 3L0 6m4.5-3H9'/%3e%3cuse xlink:href='%23b' stroke='%23ffb81c' clip-path='url(%23c)'/%3e%3c/g%3e%3cuse xlink:href='%23d' fill='none' stroke='%23007749' stroke-width='1.2'/%3e%3c/g%3e%3c/svg%3e){kind=small}
Scoring Function
Help · 22 Aug 2024, 10:33 · 16
Hello!
Thanks for bearing with us. We're pleased to share the scoring function with you. Please find it here - https://zindi.africa/competitions/smart-energy-supply-scheduling-for-green-telecom/data
Please note that -1 submissions will be removed from the leaderboard today. Moving forward you will be provided with an error message in these cases.
Good luck!
I'VE BEEN GETTING The site name or time length is wrong, please check your submission
I'VE BEEN GETTING The site name or time length is wrong, please check your submission and The strategy is unfeasible, please re-design your strategy. every time i make submision or am i the only one experiencing ths?
Hi @meganomaly THERE IS AN ERROR in the calcSingleSiteScore function (below) with potential huge impact on the leaderboard.
Indeed when the battery discharges, the function uses the discharge coefficient till the end. But the discharge coefficient (1.2) is greater than the charge coefficient (0.2). This will artificially increase the stage of charge (SOC) and WILL MAKE FEASIBLE SOME STRATEGIES THAT SHOULD NOT HAVE BEEN FEASIBLE. And it adds randomness to the evaluation because once it discharges for the first time, the charging steps are calculating with the discharge coefficients. Strategies that forced earlier discharges will be advantaged.
This is the two problematic lines of the function:
#
coe = float(config["coe"])
😎
Nice Catch. @Zindi @meganomaly
Thank you @marching_learning! We are taking a look
Thank you @marching_learning for this important catch !
this problem solved and error metric script updated
hi @meganomaly please check scoring func for grid outage plan. when grid plan is outage is true means no power will be taken from grid and when it is false the grid power is available. but in score function it takes as true is aviable and false is not avialble. for more you can check thier presentation 15:07 min https://www.youtube.com/watch?v=XuCU76W7pI0&t=906s
Huge important catch @Yisakberhanu. So I was right to insist concerning scoring function 😎. Now this is the second error in the scoring script. This makes the competition even sweeter. Couldn't wait for the scoring script update. Potentially this will increase the feasibility solutions space.
This has been affecting my submission. @meganomaly. Please when it is corrected let us know!
@meganomaly
Could you clarify if this is a bug and if yes if you are going to fix it?
I guess there's a way to check the plausibility of this, if you know that outages are either very frequent or very sparse or if you know if they are more supposed to be at night or during the day.
My bad, thank you :) !
EDIT: I don't know why my original message got deleted, in case someone else makes the same mistake as me (debunked below by marching_learning):
I had thought that there was a mistake in the scoring function, that the rated_voltage of the battery was not taken into account in the state of charge dynamics. But it's my mistake.
No in this case they got it right because of the second line:
Read it back, you'll see 😎.
curSoc += sum * coe / (4 * capacity)
@meganomaly why we multiple capacity with 4, is it right ?
Without the 1/4 it would be the SOC after 1 hour, but the time step is 15 mins, I think it is ok.
where is the test data?