Multi-dimensional dynamic visualization of spatiotemporal data for tactical analysis in table tennis

Table tennis is a high-speed racket sport that generates complex technical and tactical data, including stroke sequences, landing-point distributions, shot techniques, player names, and point outcomes. Traditional statistical charts and two-dimensional visualizations are often limited in representing multidimensional categorical relationships and the dynamic evolution of landing patterns across rallies. This study therefore aims to develop a visual analytics framework for exploratory post-match technical and tactical diagnosis in table tennis.

A multimodal visual analytics framework was developed by integrating a cluster-enhanced parallel coordinates module with a stroke-sequence-oriented spatiotemporal cube module. In the parallel coordinates module, K-Modes clustering was used to aggregate categorical stroke-event records into representative tactical profiles, thereby reducing line occlusion and revealing co-occurrence patterns among player name, stroke number, stroke position, landing position, shot technique, and point outcome. In the spatiotemporal cube module, landing points within each rally were modeled as discrete landing-point sequences, and sub-trajectory clustering was applied to identify recurring landing-path patterns and the evolution of landing hotspots across stroke sequences. Two independent match cases with reliability-checked annotations were used to evaluate the two modules according to their respective analytical objectives. In addition, five experts assessed the system using a seven-point Likert scale.

The results indicated that the cluster-enhanced parallel coordinates module improved the readability of multidimensional categorical data and supported the identification and interpretation of player-specific tactical profiles. The spatiotemporal cube module revealed recurring landing-path patterns and stroke-sequence-based changes in landing-point distributions. The expert evaluation further supported the analytical usefulness and interpretability of the proposed framework for post-match technical and tactical analysis.

By combining categorical tactical-pattern analysis with spatiotemporal landing-sequence analysis, the proposed framework provides a domain-adapted approach for examining both multidimensional technical-tactical relationships and the dynamic evolution of landing patterns in table tennis. The framework supports exploratory tactical diagnosis and offers methodological implications for the visualization and analysis of spatiotemporal data in racket sports.