In this “getting started” competition, you’ll use time-series to predict the demand forecasting on weekly basis.
Specifically, you'll build a model that more accurately predicts the weekly sales . You'll practice your machine learning skills with an approachable training dataset of price of goods, crude price, us govt incentive and all Covid related information.

Object of this problem is to predict the demand forecasting for 12 weeks based upon the previous data and sale.

Dataset contains rows (~140) and around 160 features.

Challenges you might face

1 – Feature selection and Feature engineering – small subset of variables to pick (~15-30) from the entire huge list of variables, not all variables will be relevant.

2 – Hyperparameter tuning – Model will be sensitive to Hyperparams (learning rate, epochs, num of LSTM cells, epsilon, rho, beta etc) – what would be a right combination to get minimum error