Tuesday 12pm, 15 September 2015
Higher-level Tools for Interactive Data Visualization
PhD Student - Stanford University
Data visualization has gone mainstream. From business intelligence to data-driven journalism, society has embraced the use of visualization to record, analyze, and communicate data. However, crafting effective visualizations remains difficult as it requires a cross-cutting skillset. In this talk, I will present a new stack of tools for interactive data visualization. At the foundation is Vega, a declarative visualization grammar for interactive web-based visualizations. Vega extends an existing grammar of graphics with a new grammar of interaction: interaction techniques are defined through composable declarative primitives rather than with imperative event handling callbacks. While Vega is useful in its own right, for example Wikipedia uses it to embed visualizations within articles, its declarative format is particularly amenable for programmatic generation by higher-level tools. I will demo one such system, Lyra, which provides a direct manipulation graphical interface for visualization design. Since its April 2014 alpha release, approximately 1,500 users have used Lyra each month and have found it to be an effective prototyping and teaching tool. Lyra is part of a nascent Vega ecosystem which also features Voyager, a recommendation-powered visualization browser, and packages to integrate with the iPython Notebook and RStudio. As this ecosystem matures, it holds the promise of interoperability: produce exploratory data visualizations with one tool and export it to another for further design customization, for example.
Arvind Satyanarayan is a Computer Science PhD candidate at Stanford University, where he works with Jeffrey Heer (UW) on new languages and systems for visualization design. Arvind is also an advisor at Apropose Inc., a Bay Area start-up he co-founded to build data-driven web design applications.