Swipytics

Leveraging Short-form Video Stream Metaphor in Exploratory Data Analysis

Abstract

We present Swipytics, a learnable mobile interface for exploratory data analysis (EDA) inspired by the metaphor of short-form video stream platforms. While EDA tools have become increasingly accessible to a broad and diverse audience, the cognitive burden of selecting data attributes and constructing visualizations remains a significant barrier. Several interfaces introduced post-WIMP interactions or visualization recommendations to mitigate the barrier, but they still require users to learn the interfaces by receiving verbal instructions or watching lengthy video tutorials. To bridge this gap, we leverage the metaphor of short-form video streams, offering minimal and familiar user interactions for EDA, such as swipe gestures for exploring new visualizations. New visualizations are recommended at a linear and controllable pace to alleviate the information overload. Our user study demonstrates that participants quickly achieved high levels of proficiency and confidence in EDA after a short self-learning period (a hundred seconds on average) without any verbal guidance or video tutorial.