Look into time series models (e.g., climate prediction models)

[xingjian-bai]

HOW should we treat the data differently, given it's a time-cts climite prediction problems.

Helpful posts:

• https://www.kaggle.com/competitions/widsdatathon2023/discussion/376574

• https://www.kaggle.com/competitions/widsdatathon2023/discussion/377266

• https://www.kaggle.com/competitions/widsdatathon2023/discussion/376571
• Search more on Kaggle

[YuanYuan0608]

Be careful when choosing machine learning models. https://www.kaggle.com/competitions/widsdatathon2023/discussion/377400

"I'm seeing a few people try out tree-based models in the code section and I just wanted to give a heads-up that tree-based models will introduce data leakage by using future dates to predict past dates depending on the split of the data and you wouldn't have access to this in production for deploying the model."

[YuanYuan0608]

https://www.kaggle.com/code/mpwolke/wids-2023-nn-estimator Build neural network