Dear Zindians,
We would like to apologise for the slow release of the private leaderboard. Zindi has altered the way we create and deploy error metrics, which unfortunately has caused several interrelated bugs leading to a number of delays and leaderboard changes in this competition.
A short explanation of what happened in the scoring of this competition:
We'd like to apologise for the constantly changing leaderboard and disruptions to the platform. With several fixes implemented, the leaderboard should be fair and consistent now, and we expect to have resolved this issue for future competitions.
If you have any queries about your submission, please do comment here (replies will be back online soon) or inbox me.
The top 20 users will be considered for prizes. Please look out for an email from ITU regarding what, when, and how to submit your solutions.
Thank you for your patience and understanding as we learn and grown together.
How to submit code? I did not see any place to submit the code. I also did not receive any email submitting code.
Hi mapplelife,
If you are in the top 20, you will receive an email with instructions on what to submit within 24 hours.
Hi, I am ranking 19, but I still did not receive the email. Could you please check it for me?
I'm thinking about point 3, that the users who did not make a submission selection, should not assesment\present in leaderboard or should assessed by maximum score. Many other participants who made submission selection had better scores in their other hundreds submissions. Therefore, the leaderboard should be scored on the minimum score basis from all submissions to ensure fairness and accurate representation of performance. In this case participants from point 3 would be in equal conditions with the others ..... otherwise they got the advantage over others.
This would be against Zindi's terms and conditions.
What you suggest would increase the luck factor in determing winners
Those who do not make a selection are judged by their best public score which in normal circumstance they would have selected it.
This best public score is not always the best in private leaderboard
Yes thanks, I already saw this point in the rules. But then it’s not clear why the selection factor is needed.
You still don’t see the private until the final and always choose your best result most likely from the public. Otherwise, if you choose according to your own CV score, then why do you need a leaderboard if it does not correlate with your cross-validation on train set in any way? A closed circle :))
But in real life you need to choose model variant anyway and you haven't any public score, due to test data imitate production when your model faced with new data.... So I think model choosing factor must be a strict condition, for real life imitation in competition.....
yea
Hi. Can you confirm that only people in the top 20 have to submit their codes? I keep receiving some mail without being in the top 20. I am in the top 30. Thanks
Good day. Are you saying that the top 20 must submit their codes? Despite not being in the top 20, I continue to receive some correspondence. At the 30th spot, I find myself. I am grateful. geometry dash