Nothing hurts more than this.
The effort-to-reward ratio felt like cooking a five-course meal and only getting to sniff it if you didn’t win. Hopefully, the winner this time will be generous enough to share their solution even just a write-up would go a long way. Last time (Amini Cocoa), things were a bit quiet on that front, which was a shame since we're all here to learn from each other. Maybe we can kick things off by revisiting Amini Cocoa? @Brainiac
Huge shout out to my teammate @nymfree we did great!
@Koleshjr I've been a bit tied up with other projects lately, but I’ll definitely share my solution summaries for Amini Cocoa and this comp soon — better late than never, right?
Also, big congrats to you and @nymfree — you two really brought your A-game. Well deserved! 🔥👏 Looking forward to seeing what everyone cooked up this round.
yes better late than never, Thanks
@Koleshjr 1st Place Amini Cocoa Solution
Well said — we're here to learn from one another. But how can we truly learn if we don’t understand how the winner approached the problem differently? Sharing the winning solution is essential for us to review and gain meaningful insights.
@Koleshjr, why are you not at the top yet your submission looks like it beats the top, you had 0.976, the winner had 0.975! I am just curious.
When the competition ends , you have to choose two subs to be scored in the private LB (Which is very fair btw) Makes it even more interesting. In this case we never trusted those two subs and we ended up choosing ones that performed well in the public leaderboard which were used in scoring. it's part of the game still, that the subs you choose have to be subs you trust. Given a second chance we still wouldn't have chosen the two, Maybe we were just meant to be second this time :)
That’s fair, given it had a 0.91 score on LB, and I respect the rules and the strategic aspect of choosing which submissions to trust — it's part of the game. Still, beyond the LB, many of us are here to grow and improve. That’s why seeing how the winner approached the problem differently would be incredibly valuable. Even if the final ranking doesn't change, sharing the winning solution helps the whole community level up. We all benefit from that kind of openness. I think you should share yours, the one that had the 0.976 then, since you're the winner, as per say. Thanks!
Honestly I just want to learn too. I am just a bum at using sentinel data😅
I always share my solutions , So I will definitely share once the evaluation ends. You too can share yours. I mean everyone in the top 10 should always share their solutions. Wish this was a rule, but hey we can't force people, all we can do is ask nicely 😅.
We're still waiting for you @Koleshjr.
soon, been busy travelling
Sure big man, take your time. We would be waiting.😁
@Keleshjr we have a score of 0.95 public and 0.98 private but unfortunatly @zindi didnt take it.
First congratulationssss!!! Private LB only score subs which you chose, Most likely you never chose that one
Amazing. They only choose 2 with high public LB, so if you overfit to public LB you're done!
Hello @ML_Wizzard,
Did you get this 95 score before the competition ends ? Or after it ?
Is there a special secret behind it ?
delete
Congrats on your win mhan! Can't waiting to learn about your approach once evaluation is over. Your inundata flood mapping solution was very helpful to me in setting up my image processing and classification pipeline. Super grateful for your awesome work as always!
Appreciate it bro!