Primary competition visual

GeoAI Ground-level NO2 Estimation Challenge by ITU

Helping Italy
1 000 CHF
Challenge completed 12 months ago
Prediction
804 joined
372 active
Starti
May 22, 24
Closei
Nov 15, 24
Reveali
Nov 15, 24
I have a cross validation metric of 6 but way way worst on leaderboard, what could be the issue?
Help · 27 Jul 2024, 14:24 · 4

I tried everything to handle it, 6 on validation and 16 on leaderboard

Discussion 4 answers

Could be due to overfitting

27 Jul 2024, 21:56
Upvotes 0

Hi I am fairly new to ML. I created training and validitation datasets and despite both of them scoring quite low in RMSE (around 4-7 range), they scored 10 and 11 RMSE in the public leaderboard. Are there any tips and tricks to prevent or reduce overfitting, or any indicators that quantifies how much your model has overfit?

Essayez de representer la courbe d'apprentissage pour voir un peu comment le modele se comporte et s'il faut realiser une validation croisée bien complexe , fait le

31 Jul 2024, 21:04
Upvotes 0
User avatar
Jibreel

you can try cross validation to see how is your model generality

15 Aug 2024, 10:05
Upvotes 0