Title
AWS re:Invent 2023 - Creating new analytics in sports (SPT201)
Summary
- AWS machine learning is used to analyze sports data, providing insights like predicting NHL face-offs.
- Julie Souza, head of sports for AWS global professional services, moderated a panel discussion on sports analytics evolution.
- Panelists included Andrew Reich (AWS senior sports consultant), Brant Berglund (NHL senior director of coaching and GM technology), Cassie Campbell-Pascall (broadcaster and former captain of Canada's national women's team), and Leon Lee (AWS principal cloud architect and data scientist).
- The NHL collects data using an infrared camera system, player tags, and pucks with embedded lights.
- Data is used to create analytics like shot and save analytics, face-off probability, and opportunity analysis.
- Opportunity analysis is a supervised machine learning model that predicts the likelihood of a goal based on various factors.
- AWS SageMaker is used for building and deploying machine learning models.
- The architecture for analytics involves streaming data from NHL arenas to AWS cloud services, processing it, and delivering insights to broadcasters and teams.
- The panel discussed the iterative process of model testing and feedback, the importance of storytelling in analytics, and potential future analytics explorations.
Insights
- The NHL's use of AWS services demonstrates the potential of cloud computing and machine learning in transforming sports analytics.
- The panel highlighted the importance of collaboration between data scientists and sports experts to ensure analytics are relevant and understandable.
- Real-time analytics can enhance the fan experience by providing insights during live broadcasts and potentially influencing in-game decisions.
- The discussion on future analytics, such as blue line turnovers and defensive gaps, indicates a trend towards more granular and situational data analysis in sports.
- The use of AWS SageMaker for model development underscores the growing role of specialized machine learning tools in sports analytics.
- The panel's focus on storytelling and fan engagement suggests that the ultimate goal of sports analytics is not just to collect data but to create narratives that enrich the viewing experience.