The National Basketball Association (NBA) is considered the world's premier professional basketball league for men, and it records massive amounts of game data at an unprecedented rate. However, it is difficult for the average viewer to understand and analyze the NBAʹs high-dimensional professional data, and data visualization and machine-learning techniques are difficult for the average viewer. Data visualization and machine-learning techniques can be combined to extract key information from these data, present complex data in the form of visual charts, improve the readability and operability of the data, facilitate users to observe and understand the meaning and pattern of the data, and use the data to predict the outcomes of NBA games.
This paper details a study on the regular statistics and game time-series data of NBA players and teams, combined with data visualization and machine learning techniques, to explore and predict. The NPIPVis system first uses parallel aggregate ordered hypergraph visualization (PAOHvis) to create a super dynamic chart of NBA season wins and losses and then uses iStoryline. Moreover, iStoryline was used to customize the NBA game plot visualization component, with many chart styles and user-friendly interaction methods. In addition, an integrated learning algorithm, SRR-voting, was proposed to predict NBA all-star players, and Calliope was used to draw NBA visualization data stories. NPIPVis visualizes and interprets important NBA data, which can help general users better understand and analyze NBA teams and games, as well as understand and predict all-star players.
