Implementing End-to-End Demand Forecasting with Databricks and MLflow

Discover how to implement an end-to-end demand forecasting solution for retail using Databricks and MLflow in this 36-minute conference talk. Learn how to improve efficiencies and sharpen fresh product production and delivery planning by leveraging distributed computation with Spark to train hundreds of models in parallel at various levels, including store and product. Explore the use of Delta Lake, feature store, and MLFlow to build a highly reliable ML factory. Gain insights into how this setup runs at various retailers, feeding accurate demand forecasts back to ERP systems to support production planning and delivery. Delve into the project scope, file ingestion, extra data sources, feature engineering, model selection, and granularity. Understand how to utilize parallelism, track performance, choose appropriate metrics, work with reliability buckets, and integrate results into client systems. Cover the frequency of training and scoring, as well as monitoring and alerting strategies. Get inspired to use data and AI not only for efficiency gains but also to decrease food waste in the retail industry.