Visualizing American life
DG-supported map research is charting future uses for U.S. government geospatial data
By
dg.o Communications Manager
The DG Quality Graphics projects
|
" ColorBrewer
" Conditioned Choropleth Maps
" HealthVis
" MapStats for Kids
" Quality Graphics home page
|
The popular press often lump the study of fusion, genetics, superconductors and other high-science disciplines under the impressive label, "leading edge research." They write glowingly of disciplines where researchers are exploring the boundaries of science in search of significant, but still-distant real-world benefits.
Yet Digital Government researchers are also leading the way across uncharted territory, toward more immediately achievable real-world solutions that could have equally broad impacts upon society.
DG scientists at three universities and eight government agencies are mapping - quite literally - the gap between government mandates and applied research.
Conditioned Choropleth Maps:
|
Dynamic Multivariate Representations of Statistical Data
Research conducted by Dan Carr, Alan MacEachren, Duncan MacPherson, Erik Steiner and Mark Harrower
Enlarge map | Demo site
|
Funded by the Digital Government program, these unprecedented collaborative partnerships are exploring methods for translating raw government geospatial data into concrete tools - data maps and analyis methods that can help us learn about the nation's health, habits, living conditions and family makeup.
They are also learning of the extraordinary opportunities for pure collaboration and discovery that Digital Government grants offer both the government and academic communities.
Born of an initial planning grant directed by George Mason University statistician Dan Carr, Penn State geographic information scientist Alan MacEachren and Rice University statistician David Scott, the Quality Graphics projects now encompass work by researchers at all three campuses.
The projects are exploring methods of analyzing data from the U.S. Bureau of the Census, National Agricultural Statistics Service, Bureau of Transportation Statistics, Bureau of Labor Statistics, National Center for Health Statistics, Energy Information Administration, National Cancer Institute, and the Environmental Protection Agency.
The Digital Government projects have been fruitful for the U.S. Census Bureau, says Trudy Suchan, a statistician at the bureau's Population Division.
The collaboration is developing methods for mapping census data that can help explain anomalies in raw data sets that would otherwise be difficult to analyze, she says.
"People who are looking at the inputs to these population estimates ... are really concerned about the quality of the data they're using," says Suchan. "They want to look at their data in multiple ways, so that they can look at tab sorts, run loss functions and run time series to see if there are outliers from year to year. And adding a geo-reference basis gives them another way to look at change across time."
ColorBrewer.org
|
Dr. Cynthia Brewer (CoPI) at Penn State developed the ColorBrewer application in collaboration with Mark Harrower.
Try ColorBrewer
|
For example, a sudden shift in hundreds of people from a certain community might be a confounding anomaly in raw data - but an easily explained migration caused by a military base closure when analyzed in conjunction with a georeference map, she says.
A rich demo site, designed by Dr. Mark Harrower (who just completed his Ph.D. and started as Assistant Professor of Geography at University of Wisconsin-Madison), can be found here.
Projects range from elegantly simple map-design color studies to complex explorations of methods for parsing massive, national-level, multivariate datasets.
ColorBrewer is a Penn State-based project that developed an online, Flash-based dynamic tool that allows non-experts to choose good color schemes for complex maps so that they will render clearly and can be interpreted easily, whether displayed on CRTs, laptop screens or printouts.
One of the more elaborate studies is Conditioned Choropleth Maps: Dynamic Multivariate Representations of Statistical Data, being conducted by GMU and Penn State researchers with a variety of statistical sets. The project is developing cartographic tools such as dynamic map readers that let users view datasets in "layers" upon a map so as to explore relationships between a variety of data, such as rainfall, poverty and lung cancer rates.
MapStats for Kids
|
The sub-project team, supported by the FedStats Task Force, is: Alan MacEachren, David Howard, Mark Harrower, Bonan Li, Steve Crawford, Roger Downs, Mark Gahegan, and Sven Fuhrman
Demo site
|
The tools are being integrated in many cases with GeoVISTA Studio, Penn State's Java-based visual programming environment for creating mapping applications. The suite allows researchers under other DG grants - including GIS experts such as the University of Maine's Peggy Agouris, to plug in and create new applications.
And a sub-project called MapStats for Kids provides interactive geography lessons with Flash-driven interpretations of FedStats voting data from the 2000 presidential election.
The chance to work on real-world data has been a refreshing change, and offered more of a learning experience than the usual internal research projects, says Alan MacEachren, director of Penn State's GeoVISTA Center, and principal investigator for the Penn State component of the Quality Graphics projects.
"On the typical NSF project, you sort of develop your research ideas based on whatever theoretical perspective you're working on, you convince the reviewers it's a good idea and then go ahead and work on it without too much attention to how it might be applied in the real world," MacEachren says. "In this case, it encourages you to think about what the [partner] agency needs."
HealthVis:
|
Multivariate Representation Methods for Time Series Geo-referenced Health Statistics Data
MacEachren (PI), Edsall (Co-PI), Sponsor and Collaborating Agency, National Cancer Institute
|
"We've spent a fair amount of time trying to find out what some of the problems are at each of these agencies, doing some task analysis where we go to the agency and interview some of the analysts to determine their analysis needs and presentation needs - to help the analysts at the agencies create better statistical summaries - and then helping them to develop tools that would allow them to analyze the data," he says. "It's certainly fun to see some of the research actually get into practice."
One factor has made the collaborations particularly fruitful, he says: the Census Bureau and other government partners in the project are predisposed toward data analysis and using IT methods to explore theories.
"In general, I think the collaborations work the best when there's somebody in the agency who really has an interest in the research, not just in the final application of the research," he says.
"I think this is a useful thing for other people applying for this type of grant to realize," MacEachren says. "The other thing that really seems to make a difference is in the situations where the agencies have decided to invest even small amounts of money. It doesn't have to be large amounts, it's just that on their side, that commitment to the project causes them to ask more questions, causes us to give more answers back and facilitates the exchange of ideas."
|