VisFiles: presentation techniques for time-series data

• Authors: T. Todd Elvins
• Venue: ACM SIGGRAPH Computer Graphics
• Year: 1997

This is the first in a series of columns on the subject of data visualization. I'm excited to have this opportunity because it will give me an excuse to finally learn about some subjects that I've wanted to learn about but never gave myself the time. Gordon Cameron, Computer Graphics Editor, has given me some latitude on topics so future visualization columns will deal with color selection, data visualization systems and APIs, data visualization research, information visualization, user interface issues, tips and hints, etc. If you have comments on this column or suggestions for future columns, please send me email ([e-mail]).In this issue I will give an overview of some familiar, and some probably unfamiliar, time-series data presentation techniques. Time-series data visualization is becoming an increasingly important topic as data archives grow exponentially and computing becomes ubiquitous. Some researchers claim that time-series data can be imaged using techniques similar to those applied to spatial data (since time is most often linear), and some claim that temporal data is a much different beast. For example, temporal data can have multiple strands and events, features not present in spatial data. There are also many interesting unsolved problems and data handling issues associated with temporal data:• Boolean and other mathematical operations• Data storage and temporal databases• Data query and retrieval• Data generalization (handling data at vastly different resolutions)• Data models and representation• Data presentationI have lately been working on designing a user interface for a simple geographic information system capable of manipulating and compositing time-series geo-referenced data. This project started me thinking about the different ways of visually representing temporal data. Without going into much detail on any one method, following are all the methods I could find within arms length in my office. Most of these methods allow the scientist or visualizer to present the time-series in a single image, while the final method uses an animation technique. Good techniques for visually presenting time-series data will:• Reveal spatial and temporal relationships• Uncover patterns• Show variability