The aim of the competition is to forecast daily item sales for the next 28 days
We reached top 15% (842/5558) using a stacking model of lightGBM, catboost and Prophet with a score of 0.49436 in the public leaderboard and 0.74755 in the private
link to kaggle : https://www.kaggle.com/c/m5-forecasting-accuracy/overview