For several years, technologies related to big data and AI have been exploited by the famous fast-food chain, McDonald’s. The multinational has announced that it will strengthen its use of machine learning and data science. The firm claims to have supplied more than fifteen use cases and sent around thirty models to production. According to his forecast, machine learning use is expected to increase four to five times by the end of the calendar year.
A “democratization” of machine learning within McDonald’s
As part of the Data + AI 2021 summit which took place in digital from May 24 to 28, 2021, Patrick Baginski, senior director of data science and data analysis at McDonald’s, spoke of the firm’s ambitions regarding to its use of machine learning around the world:
“We use out-of-the-box capabilities extensively and therefore we also want to continue to educate and provide learning opportunities for our teams, the marketplace, the different people who are part of the McDonald’s system. We want to implement more end-to-end automation and machine learning operations, in general […], and we want to continue to implement governance, as well as cost control measures to ensure that what we do from a business perspective continues to make sense. ”
In the course of 2020, the fast-food chain introduced the tools of the DataBricks company, known for its Delta Lake, administered MLflow and Koalas solutions.
The implementation of a process to exploit machine learning and data science
Abhi Bhatt, world director of data science and data analysis at McDonald’s, explained in his remarks how the data would be exploited by McDonald’s thanks to the DataBricks platform of tools:
“The way we do it is this: we try to bring all the data into a s3 bucket where the data lake is activated. […] which helps us to do the versioning of the data and also to build the engineering pipelines of the scalable and efficient functionalities within the platform. We not only identified the tools, the technology, we did the legal paperwork which can always be a hassle, but we also identified the use cases, built the models and we deployed them. ”
The company would have gone from a virtually non-existent use of machine learning and data science to a “production scale of machine learning operations”. Since the launch of its digital transformation strategy, McDonald’s has managed to offer more than fifteen use cases as well as thirty models in production that have been deployed in the countries where McDonald’s operates.
Machine learning should be used a lot more within the firm by the end of the year
Several machine learning-related deployment frameworks have been built for users to ensure, according to McDonald’s, appropriate security frameworks for developing and deploying machine learning models. Currently, there are 130,000 Databricks units used for approximately 27,000 hours of compute consumed, all on a monthly basis.
Automation of marketing-related activities remains a priority for McDonald’s, such as customer interaction and cross-channel performance measurement.