UNICEF Arm 2030 Vision #1: Flood Prediction in Malawi
$10,000 USD
Predict flood extent caused by storms in southern Malawi
1604 data scientists enrolled, 480 on the leaderboard
ForecastStructuredSDG13SDG15
Malawi
2 December 2019—17 May 2020
"Big unforgivable mistake that we cannot pass away"
published 22 May 2020, 10:04

Well, I guess only the lazy one hasn't noticed the scream of @FADHLOUN about disqualification of our team from this competition. I didn't want to reply to those footless accusations, but this is going too far from this point. Obviously I don't like it when one person is bringing his own suspicion to the public, blaming our team without any evidence. What I don't like more when somebody is questioning ("I'm sure that they can do more It's just a sort of camouflage..") the skills and work that have been done to the prior point.

I don't want to explain this situation once again, I just want to tell, that we have nothing to hide, and you have no right to falsely accusing our team of cheating. We've submitted our solutions to ZINDI 2 times without any delay: first place solution more than 10 days ago before the competition end, and winning solution 3 days ago. The leak that was found in the allowed sources was promptly reported to ZINDI immediately in case they were unaware.

In our solution, we didn't use anything that is disallowed of violates the rules. And again, that is why we bring it to ZINDI team. What I don't understand, why you can't just wait until ZINDI posts a final leaderboard, I'm sure they will do their job in the best way.

Obviously I'm a new guy here. You have a great community with practical competitions and knowledgable discussions, but the toxicity of people eager for victory is extremely discouraging.

I'm very much looking forward to your solution!!

I agree Zindi should be a completely healthy platform to participate in, and if there was a leak in the allowed data sources itself (I am not aware about the exact reason), which is quite possible, it's none of you or your team's mistake, you utilized the data properly. I welcome you warmly to this Zindi community, and hope you continue participating and winning more such competitions.

Just keep up the great work @yuriimalna. Its absurd that such things are happening but hope @zindi finds ways of mitigating this propaganda.

Пусть пост будет на русском, потому что меня тоже бомбануло с этого парня, который попросту рвет себе задницу из-за места в ЛБ, причем не просто рвет, еще и орет на всех, говоря что мол "Не может этого быть, просто нереально, лучше нашего быть не должно, это жульничество". Алчный и безтактный тип....

Data leaks are part of the competitions, there are sooo many kaggle competitions that have leaks and people do not get disqualified ! If you extracted features from datasets allowed by Zindi. It must be fine !

Well this will be my last response to this particular subject .. first of all we don't know what kind of data you did use . Even when I asked you if Zindi deleted that data of not you said no .. 99% of the competitors have been working of this particular challenge since months and I didn't blame any of them without proofs .

If you were right in the beginning why would you report that to Zindi ? It's an allowed data and you have the right to use it . Again having almost the target of the test set gives you all hidden secrets of the data and the problem is over , CV is Guaranteed ,validation was one of the biggest problems in this challenge and a lot of competitors have been downranked since they didn't predict well that 50% of the data and it was the case of team "Overfitters" and "MightyGorgons" and many others.

CV was unstable, we did use a lot of tricks but given that instability and the limited number of submissions per day we were struggling to decide whether we are in the right path or not..BUT if your case man you can even approximate your private score with that 0.028 submission Hahaha what are you talking about? Since you got that score you didn't make any submission until the last day .. If you didn't have that submission how will you know that you are not overfitting the public leaderboard?

Please RESPECT our minds

"Please RESPECT our minds" I respect everybody, the exact opposite I can see from you.

"If you were right in the beginning why would you report that to Zindi ? It's an allowed data and you have the right to use it" - you're right, due to the competition rules we could use it, but it would be stupid. It was reported because for the reasons I desribed long time ago.

"Since you got that score you didn't make any submission until the last day .. If you didn't have that submission how will you know that you are not overfitting the public leaderboard?" Well, you're wrong. The winning submission was the one 3 hours before the competition end. And I have both good scoring and bad scoring models as my last attempts.

"CV was unstable, we did use a lot of tricks but given that instability and the limited number of submissions per day" well, I was in this situation as well l.

Why don't you tell everybody how you were writing me privatelly in Linkedin and asking what was the raster producing that score? :D You're just ridicolous.

"You're just ridicolous." Thank you , being honest these days would end to this ;)

I can post our linkedin discussion if you want where I did ask you if Zindi deleted that particular Raster and I suggested to you if It was that flood frequency layer that give that score cause I had doubt about that since it gived me a boost in my models and then the discussion ended there . I was surprised that Zindi didn't take any action in order to delete it . I wouldn't use that raster since I thought Zindi will Prevent that layer , Again having a score around 0.028 would lead me to disqualification.

Thank you again for this responses , You just responded the questions you wanted to and again sinking yourself LOL . I saw that unconfortable situation that you were in and how you were frustrated about the number of submissions and CV .. last 6 days jumped to first position then practically no submissions till the final day and then back with few ones hahahahaha , I sware that's a too funny situation that you're trying to defend . You made my day hahaha Thank you .

To end Why didn't you speak about that 0.028 submission file? Aaah okay I understand it was your black horse for this competition . If It was a fair challenge every competitor would have that particular file .. But in the case it's not we saw the results .. the ones who had that file are first and other are after them !!!

It's pointpess to have a discussion with you. Again, read the last paragraph of my heading post about the toxity, maybe you'll understand something.