Visualization of Multivariate Spatiotemporal Data


Project Start Date: Jul 1, 2012
Funding: Member Funded
Project Tags: , ,

Project Summary

Data analysis is a critical component of every business decision making and management process. A wide variety of systems are used to provide a quick and easy to understand overview of important events within the business. A widely used approach for managing these tasks is web-based dashboard systems. While these are effective at presenting information, they are very limited in terms of interactivity and exploration. They provide a good way to show what is going on, but they fall short of being able to answer why.

The primary objective of this project was to evaluate the feasibility of providing interactive visualization capabilities for large-scale data (millions of records) within a web browser environment to potentially enable seamless integration into/extension of existing dashboard systems.

To evaluate the basic capabilities and limitations of a Javascript-based web browser environment we built and benchmarked a set of basic visualizations using industry-standard techniques. These turned out to be severely limited in terms of scalability, making them totally unsuitable for use on realistic datasets.

To avoid these limitations we designed and prototyped a totally different approach that moves as much of the visualization data and process as possible into the graphics hardware using WebGL. This includes not only the rendering of the graphical primitives but also the data storage, data selection and primitive generation steps.

The developed feasibility demonstrator can handle datasets with more than a million records at interactive rates, proving that it is indeed feasible to use a web browser environment for interactive, real-time information visualization tasks. This fact provides a great motivation for continuing the work and exploring more complete and flexible data representations and visualization methods.