Designing Contextual XR Visualizations for Sports Analytics

In the modern world, the ubiquity and relevance of data in our daily lives have amplified the need for more accessible data visualization and interaction methods. Traditional data visualization techniques, primarily tailored for structured environments, struggle to address the dynamic challenges and cognitive demands users face in real-life settings. This challenge is particularly evident in the realm of sports analytics, where the swift pace and physical demands necessitate more immediate and flexible visualization tools. Contextual XR Visualizations, leveraging the advancements in Extended Reality (XR) technologies, offer a promising solution by seamlessly integrating data into physical environments. These visualizations enable the presentation of relevant data within its physical contexts, thereby enhancing the accuracy and comprehensibility of dynamic spatial data. Yet, there is a notable gap in applying these visualizations effectively in real-world scenarios, particularly those that require an understanding of changing environments and user needs. This dissertation explores the design of contextual XR visualizations for sports analytics through four projects, addressing the challenges of visualizing spatial data under dynamic physical contexts and user needs. The first study investigates the design of situated visual feedback for basketball free-throw training, showing benefits in motion guidance when presenting 3D spatial data in physical space. The second study explores a design framework for labeling dynamic spatial objects in XR, showing that close mapping of labels and spatial features can significantly enhance spatial searching tasks. The third study addresses designing visualizations directly embedded into dynamic scenes for real-time analysis in basketball games, enabling interactive exploration and analysis of evolving spatial data. Lastly, the fourth study constructs an immersive analytic system for spatiotemporal data analysis in badminton video, aiding high-performance coaches effectively comprehend complex data patterns over time and space. By integrating data with its physical context in real-world sports applications, contextual XR visualizations extend the usefulness and engagement of data analysis beyond traditional desktop environments. This thesis pioneers a novel paradigm in the design of contextual XR visualizations for sports analytics, steering data visualization research towards solutions that are accessible, engaging, and con