NSF studies ways to show complex data
- By Patricia Daukantas
- Sep 11, 2002
ColorBrewer demonstrates how different color schemes highlight numerically ordered or qualitative data sets.
The National Science Foundation's Digital Government Program is devising tools that visually represent complex statistical data sets.
Alan MacEachren, a geography professor at Pennsylvania State University and one of the project's leaders, is working with colleagues at George Mason University in Virginia and Rice University in Texas.
Eight agencies are participating: the Bureau of Labor Statistics, Bureau of Transportation Statistics, Census Bureau, Energy Information Administration, Environmental Protection Agency, National Agricultural Statistics Service, National Cancer Institute and National Center for Health Statistics.
'Graphics do a good job of summarizing complex information if they're designed well,' MacEachren said. One tool developed with Quality Graphics funding is ColorBrewer, which helps users select the best color schemes for overlaying their data on maps.
Geographers Cynthia Brewer of Penn State and Mark Harrower of the University of Wisconsin used Flash from Macromedia Inc. of San Francisco to build the online tool, available at www.colorbrewer.org.Blue for rivers
The site's schematic map shows users what color blends best emphasize different types of data'for example, whether overlays of highways or waterways would be visible with a chosen scheme.
ColorBrewer advises users whether the color scheme they've chosen is suitable for black-and-white photocopying, displaying through an LCD projector or viewing by color-blind people. About 8 percent of U.S. residents have some color perception problem, MacEachren said.
Last year, Brewer worked with Census Bureau statistician Trudy Suchan to create a Census 2000 map atlas that all people can view.
Another project by Penn State and George Mason researchers is about conditioned choropleth maps.
A choropleth map shows data values by filling in state or county areas with solid colors, MacEachren said. Such a map of a state could correlate lung cancer rates with per-capita income.
Other Quality Graphics tools under development are HealthVis, an application for cross-referencing geographically linked data sets, and MapStats for Kids, a demonstration site at the FedStats federal statistics gateway, www.fedstats.gov
HealthVis started as a proof-of-concept application written in ArcView from Environmental Systems Research Institute of Redlands, Calif. MacEachren said it is being rebuilt as a Java application using GeoVISTA Studio tools developed by Penn State's Geographic Visualization Science, Technology and Applications Center.
Porting HealthVis to Java will make it accessible to users who do not have an ArcView license. It's also much easier to create Web applications in Java than in ArcView, MacEachren said.
B. Sue Bell, a mathematical statistician with the National Cancer Institute's Division of Cancer Control and Population Sciences, is working on HealthVis and another Quality Graphics effort called the State Cancer Profiles Project.
The profiles show policy-makers how reliable various cancer-related data sets are by representing margins of error as colored bars. Color codes link the data points to thumbnail maps of states and counties that the data points represent.
The State Cancer Profiles Project does not have a public Web site yet, Bell said. Researchers at George Mason developed the initial version of the application, and now NCI contractors are cleaning up the code. She said she hopes the public site will go live in December.
'It's helping us get beyond the usual government tables to present information in a more useful way,' Bell said.