Web-based Scalable Visual Exploration of Large Multidimensional Data Using Human-in-the-Loop Edge Bundling in Parallel Coordinates
Peer reviewed, Journal article
Published version
View/ Open
Date
2020Metadata
Show full item recordCollections
Original version
CEUR Workshop Proceedings. 2020, 2578 .Abstract
Visual clutter and overplotting are the main challenges for visualizing large multidimensional data in parallel coordinates, which greatly hampers the recognition of patterns in the data. Although many automatic clustering and edge-bundling methods have been used in parallel coordinates to reduce visual clutter and overplotting, a scalable, transparent, and interactive approach that allows analysts to interact with large data and generate interpretable results of visualization in real time is lacking. To solve this problem, we propose an approach, human-in-the-loop edge bundling, to visually explore and interpret large multidimensional data in parallel coordinates. This approach combines data binning-based clustering and density-based con uent drawing, which reduces much data processing time and rendering time. It provides novel interactions, such as splitting, adjusting, and merging clusters, to integrate human judgment into the edge-bundling process. These interactions make the underlying clustering transparent to users, which allow users to generate interpretable visualization without complex data clustering. The scalability of our approach was evaluated through experiments on several large datasets. The results show that our approach is scalable for large multidimensional data, which supports real-time interactions on millions of data items in web browsers without hardwareaccelerated rendering and big data infrastructure-based data processing. We used a case study to highlight the e ectiveness of our approach. The results show that our approach provides an interpretable way of visually exploring large multidimensional data in parallel coordinates.