This project demonstrates a complete, end-to-end machine learning workflow for time-series forecasting using the Snowflake ML library. The goal is to predict daily Chicago bus ridership by training, deploying, and managing an XGBoost model entirely within the Snowflake ecosystem.
This project demonstrates a complete, end-to-end MLOps workflow for time-series forecasting, managed entirely within the Snowflake ecosystem. It simplifies the machine learning lifecycle by handling secure data connection, distributed feature engineering, scalable model training, and deployment as a service. The solution streamlines development by leveraging the Snowflake ML library, Feature Store, and Model Registry for a governed, reproducible, and efficient framework. By deploying the final model to Snowpark Container Services (SPCS), it enables real-time forecasting through a single, scalable platform.
For prerequisites, environment setup, step-by-step guide and instructions, please refer to the QuickStart Guide.