Landslide Prevention and Innovation Challenge
'/%3e%3cuse xlink:href='%23a' transform='rotate(144 450 300)'/%3e%3cuse xlink:href='%23a' transform='rotate(216 450 300)'/%3e%3cuse xlink:href='%23a' transform='rotate(288 450 300)'/%3e%3c/svg%3e)
Helping Hong Kong
3 000 Zindi Points
Challenge completed ~3 years ago
Classification
571 joined
172 active
Start
Aug 15, 22
Close
Sep 30, 22
Reveal
Sep 30, 22
'%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}
GBM or CNN
Help · 3 Oct 2022, 08:13 · 2
Hi, nice comp, glad it is over.
Congratulations to the winners. If I remember correctly, Babawale was leading for a long time (and by far) so it is good that he won in the end - congrats @Babawale_Sodiq
I've seen a few published solutions, all GBM based. I tried GBM, but, given spatial nature of data, believed this is begging for a CNN solution. Sure, lots more tuning and struggle than GBM, and lower score :-(, but the 5x5 grid sort of suggests CNN with channels being the different features.
Anyhow, any others tried CNN as well? And any comments on which to use here?
I tried CNN as well but I got terrible result ( i am not expert in NN/CNN, just tried to use some examples).
Then I switched to GBM with heavy feature engineering. I think made naive mistake and ovefit my model based on public leadeboard and lack of stable CV results compare to the result of board.
I will share same model with less overfitting soon.
Congrats for the winner and thanks for sharings and launching this challenge.
Thanks @skaak, I only used GBM models. You can find my solution here : https://zindi.africa/competitions/landslide-prevention-and-innovation-challenge/discussions/13460.