Source: Kaggle Tabular Playground Series Sep 2022
Time series forecasting with multiple paradigm shifts! While the 'proper' way of doing this assumes modeling total sales and product proportions separately, I've decided to try and implement an indicator instead (the test set is separate and large so you can't just make it continuous of course).
Required units:
- Pandas
- Scikit-learn
- Scikit-optimize
- XGBoost
- statsmodels
- NumPy
- SciPy
- Matplotlib
- Seaborn