|
These are examples of applications we've created using GeoVISTA Studio
in support of our research efforts with cancer data. We encourage
you to try building and using our tools. Your feedback to us is
extremely valuable and we're always interested in new ideas and
potential modifications.
ESTAT - Exploratory Spatio-Temporal Analysis
Toolkit - Since 2003, the Geographic Visualization, Science,
Technology, and Applications (GeoVISTA) Center at The Pennsylvania
State University has focused evaluation efforts on a long term project
to develop an interactive exploratory toolkit to support cancer
epidemiology. The Exploratory Spatio-Temporal Analysis Toolkit (ESTAT)
is based on the open-source codeless programming environment GeoVISTA
Studio. GeoVISTA Studio is a Java-based environment designed to
provide technically adept users with the ability to create their
own customized geographic visualization applications.
ESTAT features a scatterplot, bivariate map, parallel coordinate
plot, and time series graph. Each of these tools is linked to the
others so that brushing and selection are instantly coordinated.
The ESTAT toolkit is available for download with sample datasets
and tutorials on this page, and additional materials are located
on at http://www.geovista.psu.edu/ESTAT.
Building ESTAT in GeoVISTA Studio:
How to
Build ESTAT In GeoVISTA Studio
Sample Dataset
Building ESTAT in GeoVISTA Studio
(.avi video clip)
Description: This clip illustrates the steps taken to build ESTAT
in GeoVISTA Studio. The process starts by adding 6 Java Beans
to the Studio design box: one bivariate map, one scatterplot,
two parallel coordinate plots (one used as a PCP and the other
used as a time-series graph), one data loader, and one coordinator.
Then, beans are wired together so that all of the graphic components
support coordinated user action and one of the PCPs works as a
time series plot (by being linked to the data loader's time series
import hook, rather than directly to the coordinator).
Using ESTAT to Explore Cancer Data:
|
Video Tutorials
|
| Title |
Format |
| ESTAT_map-scatterplot-fish |
.avi (trouble? try this) |
| Description: This clip focuses on exploration
of bivariate data relationships using the bivariate map and
scatterplot tools. The clip begins by illustrating mouse-over
ID of places in the scatterplot, then illustrates linked brushing
from the scatterplot to the map and PCP, then from the map to
the other components. Then, the excentric labeling component
is illustrated (this is a free component obtained from http://www.cs.umd.edu/hcil/excentric/).
Finally, a fisheye lens applied to the map is demonstrated.
This component is from an open source tookkit by Jean-Daniel
Fekete called The InfoVis Toolkit, available from: http://www.lri.fr/~fekete/ |
| ESTAT_stomach-benzine_extreme |
.avi (trouble? try this) |
| Description: This clip emphasizes the use
of brushing on the PCP. The variables depicted are mercury and
benzene emissions, stomach cancer, and per capita income. The
first scene shows that Los Angeles county has the highest benzene
emissions (and third highest mercury emissions). The interaction
demonstrated shows a two step selection. First, those counties
with benzene emission above the interquartile range are brushed.
Then, all the remaining counties with stomach cancer mortality
rates within or below the interquartile range are removed from
the display. The counties that remain are those that are both
high in stomach cancer mortality and high in bezine emission.
The map illustrates that these counties are concentrated in
the northeast industrial belt and in the west. The preponderance
of dark gray counties indicates that the counties are also high
in mercury emissions. |
| ESTAT_box_state_extremes_Br |
.avi (trouble? try this) |
| Description: This clip is also focused
on the PCP. The initial scene shows that the county with highest
male stomach cancer mortality for the time period is Petroleum,
MN. The rest of the video clip illustrates the use of box plots,
summary lines (one for each state), and axis focusing and rescaling.
The tools are used to identify states in the U.S. having stomach
cancer mortality rates for men that are (a) lower than the interquartile
range as depicted on the box plot for stomach cancer (mostly
in the southeast) and (b) above the interquartile range (mostly
in the northeast and upper mid-west/great plains). The map and
scatterplot depict the relation between stomach cancer mortality
and mercury emissions aggregated by county. When the PCP is
focused in on the high stomach cancer mortality counties, the
bivariate map illustrates that those in the northeast are also
high in mercury emission (indicated by the dark gray that represents
high values on both variables) while the mid-west/great plains
counties do not exhibit this relationship. |
| ESTAT_Lung_Cancer_Gender_investigation |
Flash .swf |
| Description: This video demonstration illustrates
the visual analysis of a multivariate dataset using ESTAT. The
dataset explored here is made up of lung cancer mortality counts
and rates and a variety of socioeconomic, environmental, and
educational variables, all aggregated to the county level across
the continental United States. The demonstration shows how visualization
and linked tools in ESTAT can be used to quickly and broadly
to investigate the dataset for geographic variations in any
of the variables. It then uses, as an example, differences in
mortality by gender to illustrate a deeper multivariate analysis
of the data. The analysis reveals some interesting associations
between male lung cancer mortality and other variables that
generate deeper questions and warrant further analysis with
ESTAT, and with traditional epidemiological studies. |
|
|
Visual Inquiry Toolkit
The Visual Inquiry Toolkit integrates visual, computational, and
cartographic methods to enable human knowledge and judgment to be
coupled productively with computational methods for incrementally
searching patterns. It was initially developed and applied to dataset
on U.S. industry and sales, which was provided as the benchmark
data set for the 2005 IEEE Information Visualization Contest. Our
Visual Inquiry Toolkit entry took first place. We are now generalizing
the methods for application to cancer incidence and covariate data.
Please check back for updates in the summer of 2006.
Read a summary paper of the winning entry here
(pdf file)
Watch a video demonstration of the winning
entry here
(42 MB)
|
| Data Mining - Interactive Feature
Selection and Cluster Analysis
Data Mining
in GeoVISTA Studio
Sample Dataset
Video Demos - in camtasia .avi
format
Data Mining in GeoVISTA Studio
Narrative for this
video
|
| Matrices - Multiform
Linked Scatterplot and Bivariate Map Matrices
Assembling
Matrices in GeoVISTA Studio
Using
Matrices
Sample Dataset
|
|
Want to use your own data? Read our set of instructions here.
|
| Questions or comments?
Email us at arobinson@psu.edu |
