Primary competition visual

AI4D Africa’s Anglophone Research Lab Tanzania Tourism Classification Challenge

Helping Tanzania, United Republic of
$1 000 USD
Challenge completed over 3 years ago
Classification
506 joined
180 active
Starti
Jun 01, 22
Closei
Jul 01, 22
Reveali
Jul 01, 22
User avatar
Koleshjr
Multimedia university of kenya
Cross validation
Platform · 1 Jul 2022, 15:26 · 8

Always trust your local cross validation they said. Congratulations to everyone who took part in this competition. It was a great learning experience.

Discussion 8 answers

@Koleshjr Congratulations! What type of models and cross-validation techniques did you use?

1 Jul 2022, 15:39
Upvotes 0
User avatar
Koleshjr
Multimedia university of kenya

One catboost + stratifiedKFold on the target

User avatar
Koleshjr
Multimedia university of kenya

And thank you :)

Great approach! Thank you!

User avatar
University of benin

Lol my Local CV gave me my beat score even thou it was good enough

Congrats 👏

A great learning experience

1 Jul 2022, 16:20
Upvotes 0
User avatar
msamwelmollel
University of Glasgow

Always, indeed, trust your local CV. I had one with 1.028 in the private score, and I didn't choose it since it performed poorly on the public leaderboard, It has 1.030 as I am seeing it now. Congratulations to the winner. Let's meet next time.

1 Jul 2022, 16:58
Upvotes 1
User avatar
University of benin

Yeah congrats on your winning 🏅

Hope to learn about your approach

User avatar
Koleshjr
Multimedia university of kenya

Wow thats amazing. What CV strategy did you use ? Because i never passed 1.050 in my local validation which was (stratifiedKFold)