This project aims to develop a scalable and efficient system for forecasting store-level sales for individual items using Prophet and PySpark. The system will provide insights into future demand, which allows for better overall business planning.
This project uses Facebook's Prophet library to build individual sales forecasting models for each combination of 5 stores and 50 items over a 5-year period historical dataset. PySpark is leveraged to handle the large data volume and to faciliate parallel model training for enhanced efficiency.