Waltzboard: Multi-Criteria Automated Dashboard Design for Exploratory Analysis

• Published in: IEEE Pacific Visualization Symposium pp. 258 - 268
• Main Authors: Choi, Jiwon, Jo, Jaemin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 22.04.2025
• Subjects: Adaptation models; Analytical models; Automated Data Analysis; Benchmark testing; Dashboard Design; Data analysis; Data models; Data visualization; Exploratory Data Analysis; Heuristic algorithms; Surveys; Training
• ISSN: 2165-8773
• DOI: 10.1109/PacificVis64226.2025.00032

We present Waltzboard, an automated dashboard design system for exploratory data analysis. Despite the benefit of dashboards, which provide a glanceable overview of data, previous dashboard design systems often require precomputation, such as training deep-learning models, and do not adapt effectively to changes in the user's intent during data analysis, hindering quick and flexible data exploration. To overcome these challenges, we introduce a dashboard evaluation framework that quantifies how a dashboard describes data in terms of five key measures: Specificity, Interestingness, Diversity, Coverage, and Parsimony. We then present a three-phase search algorithm designed to efficiently explore dashboard designs without the need for precomputation. Finally, we present a user interface that allows the user to dynamically build their own intent and reason for the design process. The result of our performance benchmark and user study demonstrates that Waltzboard not only designs a more effective dashboard within seconds but also supports flexible exploratory data analysis to meet diverse analytic needs.