Congratulations to the top finishers. I am very curious about 0.26+ public lb score solutions and 0.25+ on public, I couldn't achieve it.

Was there any interesting feature at play? Did you try any recommendation specific algorithms or customer similarity-based approaches? or is it just a binary classification model? I am looking forward to your approach / solution / code.

Thank you very much Zindi and Akeed for bringing a very interesting competition for us.