The goal of the challenge was to predict the returns vs. bitcoin of clusters of cryptoassets.
My global approach was to implement a stacked regression model, containing three base models: XGBoost, RandomForest and LightGBM regressors. For this challenge, I have achieved the 11th place out of the 51 participants.
-The code I made for this challenge in a Jupyter Notebook (.ipynb) format.
-A PDF explaining my whole approach of the problem.
Unfortunately, I'm not allowed to share the dataset, but you can download it if you register on the Data Challenge website on this following link: https://challengedata.ens.fr/participants/challenges/71/
Here is the link of the leaderboard: https://challengedata.ens.fr/participants/challenges/71/ranking/public
Username: VictorHoffmann