Exploring Multivariate Data Streams Using Windowing and Sampling Strategies

The analysis of data streams has become quite important in recent years, and is being studied intensively in fields such as database management and data mining. However, to date few researchers in data and information visualization have investigated the visual analytics of streaming data. Although streaming data is similar to time-series data, its large-scale and unbounded characteristics make regular temporal data visualization techniques not effective. In this paper, we propose a framework to visualize multivariate data streams via a combination of windowing and sampling strategies. In order to help users observe how data patterns change over time, we display not only the current sliding window but also abstractions of past data in which users are interested. Uniform sampling is applied within a single sliding window to help reduce visual clutter as well as preserve the data patterns. However, we allow different windows to have different sampling ratios to reflect how interested the user is in the contents. To achieve this functionality, we propose to use a DOI (degree of interest) function to represent users' interest for the data in a particular sliding window. In order to visually Copyright is held by the author/owner(s). CHI 2008, April 5 – April 10, 2008, Florence, Italy ACM 978-1-60558-012-8/08/04. Zaixian Xie Worcester Polytechnic Institute 100 Institute Road Worcester, MA 01609 USA [email protected] Matthew Ward Worcester Polytechnic Institute 100 Institute Road Worcester, MA 01609 USA [email protected] Elke Rundensteiner Worcester Polytechnic Institute 100 Institute Road Worcester, MA 01609 USA [email protected]