Client Description

Restaurant Services, Inc. (RSI) is the supply chain manager for the BURGER KING® system and the exclusive purchasing agent for the vast majority of products and distribution services used by RSI’s members – BURGER KING® restaurant owners in the United States. Burger King has restaurant locations throughout the US and more than 100 other countries.

Challenge

RSI needed a scalable cloud based machine learning forecasting engine that could be trained to give insights to promotional demand interests.

Solution

The scope of the Demand Pomo Forecasting project was to create daily forecasts at restaurant level for each of the remaining days in a promotional period for limited time offer promotional products. This was achieved by deploying an AWS infrastructure including AWS CloudFront, AWS RDS, AWS SNS, and AWS Elastic Beanstalk that allowed clients to access and run forecasting models on a new AI/ML engine. The forecasts were to be made for all restaurants participating in the promotion and would use historical promotional sales of the particular promotional product where available for training the forecasting model. In the cases cold-start cases where the promotional product had no history, proxy promotional products would be selected for training the forecasting model.

Benefits

• RSI’s machine learning platform can now support any workload regardless of volume, velocity, and variety of data
• Business users have more trust in their data
• RSI is now exploring ways to leverage these data assets outside the scope of traditional restaurant supply & demand models
• Scaled seamlessly with little overhead to support a 70% increase in workload volume
• Built a custom ML model that runs at a target less than 20% error for all products across all stores