Visualizing changes over time in datasets using dynamic
hierarchies
John Alexis Guerra Gómez @duto_guerra
john.guerra@gmail.com
Michael L. Pack, Catherine Plaisant and Ben Shneiderman
Twitter: @duto_guerra
http://treeversity.cattlab.umd.edu
May. 22th 2013 HCIL Symposium
Human-Computer Interaction Lab &
Center For Advanced Transportation
Computer Science Department
University of Maryland
Trees are everywhere…
http://www.ncaa.com
A speculatively rooted tree for rRNA genes, showing major
branches Bacteria,Archaea, and Eucaryote.
http://en.wikipedia.org/wiki/Phylogenetic_tree
…and we are really good with them
TreeMaps [Johnson & Shneiderman 1991]
Visualizing
Storing, Computing, etc…
SpaceTree [Plaisant, Grosjean & Bederson 1998]
Navigating
Comparing trees is more complicated!
What about comparing them?
DEFINITIONS
Tree
• Set of nodes and links that express the parentto-child relationship.
• Where each node:
– Is uniquely labeled in the tree.
– Contains one or more numeric variables with
values over time.
– Contains one or more categorical attributes that
may have more than one value.
Types of Trees
Fixed
Mixed
Dynamic
State
Gender
Gender
County
Ethnicity
State
City
Age range
City
Types of Tree Comparison
Addressed by related work
Node value changes
Partially addressed by related work
Addressed by TreeVersity
No
Yes
No
Topology
changes
Yes
Leaf values
(aggregated trees)
Values in all nodes
Comparing two trees using node values and created and removed nodes
HOW TO COMPARE TWO TREES?
The StemView
The StemView
Actual Change (Color)
Direction
Relative Change (Height)
Ending Value (Width)
Created
Node
(black border
)
Removed
node
Þ
Acknowledgements
• Fulbright Science and Technology Scholarship
• Center for Integrated Transportation Systems
Management (a Tier 1 Transportation Center at
the University of Maryland)
• Center for Advanced Transportation Technology
Laboratory (CATT Lab)
• Audra Buck-Coleman
• All of our domain experts
Thank you!
StemView: shows 5 characteristics of
change in all the nodes of a tree
Reporting tool: helps users
navigating changes in the tree
http://treeversity.cattlab.umd.edu
John Alexis Guerra Gómez
@duto_guerra
jguerrag@cs.umd.edu
• ww
BACKUP SLIDES
My thesis in one minute
SUMMARY
Visualization Technique: the Bullet
Relative Change (Height)
Direction of
change
Created
node
Actual Change (Color)
Removed
node
“But, what if we want to see the changes in the
budget for the last 20 years?”
Comparing one tree over multiple points in time with node values and
created and removed nodes
HOW TO FIND WHAT CHANGED ON
ONE TREE OVER TIME
13 case studies with 9 partners
13 CASE STUDIES THAT REVEALED
STRENGTHS & WEAKNESSES
Organization Case Study
MILCS
Stage
Driving
Mode
DOT
Airlines Budgets
Early
Chauffeur
1
OMB
US. Federal Budget
Early
Chauffeur
1&2
DOT
Early
Chauffeur
1&2
52,135 8,012
674 (2 Levels)
20
1 Dynamic
DOT
TRB Publications
Nat. Trans. Library
Publications
Early
Chauffeur
1&2
38,351
374
294 (3 Levels)
10
1 Dynamic
DOT
Passengers flying in the US
Early
Chauffeur
1&2
65,534
162 4,194 (3 Levels)
4
1 Mixed
NCI
National Cancer Institute
Early
Chauffeur
2
1,716
101 (3 Levels)
3
3 Dynamic
FDA
FDA Drug Adverse Effects
Mature
Chauffeur
2
2,964
5 1,614 (4 Levels)
4
4 Fixed
UMD
UMD Budget
Early
Chauffeur
2
16,332
5 1,296 (3 levels)
6
1 Mixed
UMD Bursar
UMD Students Information
Mature
Chauffeur
2
227,158
715 (5 Levels)
219
3 Mixed
eBay
Early
User-driven
2
63,098
4 5,443 (4 Levels)
6
2 Fixed
Early
User-driven
2
96,205
286 (3 Levels)
7
4 Mixed
IDB
eBay Product Sales Data
Transportation Bottleneck
Data
Imports and Exports
in the Americas
Early
User-driven
2
119,741
19 3,766 (4 Levels)
5
1 Dynamic
DUTO
Blind Students in Colombia
Mature
User-driven
2
33,802
4 1,098 (3 Levels)
21
CATT Lab
TreeVersity Data
Version
Size
Time Example
Points Tree Size
Number Number Type of
Attribs. Vars.
Tree
216
187 (2 Levels)
3
1 Dynamic
4,845
56 1,393 (4 Levels)
7
1 Mixed
13
5
24
1 Mixed
Tree Comparison
Type
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Type 2: non
aggregated +
same topology
Type 4: non
aggregated +
different topology
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Type 1: aggregated +
same topology
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Type 3: aggregated +
different topology
Exit questionnaire
q1: For this particular case study
TreeVersity was:
q2: In general the tool is likely to be:
q3: Did the reporting tool help direct
your exploration?
q4: Did you find the StemView
comprehensible?
q5: Did you find the Bullet
comprehensible?
q6: Would you like to continue
working with Treeversity?
q7: Would you be willing to install and
use Treeversity on your own?
q8: How does this compare to your
original expectations before starting
with the tool.
“The work I can achieve with TreeVersity is equivalent to
years of analysis with our previous methods”
“TreeVersity easily showed us multi-years of the data in a graphical form at once
without having to download the data into excel and create graphics. We can do that
analysis now, but TreeVersity was much quicker.”
“I've spent many hours in color coded Excel files expressing changes at query and
category levels. The visualization and what changed views made this process much
more efficient and easier to share.”
“Numerous examples of obvious, but heretofore unrevealed patterns were discovered.”
“We use Hyperion and excel for analysis. However, TreeVersity made
it much easier to see multi-year trends.”
WRAPPING UP
Design of the Bullet + Implementation on TreeVersity
Design of the StemView + Implementation on
TreeVersity2. Design and implementation of the
Reporting Tool.
13 Case Studies that revealed strengths & weaknesses
Publications
Published
•
K. Wongsuphasawat, J. A. Guerra Gómez, C. Plaisant, T. D. Wang, M. Taieb-Maimon, and B. Shneiderman, “LifeFlow: visualizing an overview of
event sequences,” in Proceedings of the 2011 annual conference on Human factors in computing systems - CHI ’11, 2011, p. 1747.
•
J. A. Guerra Gómez, K. Wongsuphasawat, T. D. Wang, M. L. Pack, and C. Plaisant, “Analyzing incident management event sequences with
interactive visualization,” in Proceedings of the Transportation Research Board 90th annual meeting, The National Academies, 2011.
•
J. A. Guerra Gómez, A. Buck-Coleman, C. Plaisant, and B. Shneiderman, “TreeVersity: Comparing tree structures by topology and node’s
attributes differences,” in Visual Analytics Science and Technology (VAST), 2011 IEEE Conference on, 2011.
•
J. A. Guerra Gómez, A. Buck-Coleman, C. Plaisant, and B. Shneiderman, “TreeVersity: Visualizing Hierarchal Data for Value with Topology
Changes,” in Proceedings of the Digital Research Society 2012: Bangkok Vol 2, 2012, no. July, pp. 640–653.
•
J. A. Guerra Gómez, A. Buck-Coleman, M. L. Pack, C. Plaisant, and B. Shneiderman, “TreeVersity: Interactive Visualizations for Comparing
Hierarchical Data Sets,” in Proceedings of the 2013 Transportation Research Board Annual Meeting, 2013. Recognized with the Greg
Herrington Award for Excellence in Visualization Research by the TRB Visualization in Transportation Committee.
•
J. A. Guerra Gómez, M. L. Pack, C. Plaisant, and B. Shneiderman, “Visualizing changes over time in datasets using dynamic hierarchies,” HCIL
Tech Report 2013-06, 2013. [Online]. Available: http://hcil2.cs.umd.edu/trs/2013-06/2013-06.pdf. (Under review VisWeek 2013)
Tech Reports
•
J. A. Guerra Gómez, “MySocialTree: Browsing the Facebook Feed Using Hierarchies,” HCIL Tech Report 2011-29, Dec-2011. [Online]. Available:
http://hcil.cs.umd.edu/trs/2011-29/2011-29.pdf.
•
J. A. Guerra Gómez, A. Buck-Coleman, C. Plaisant, and B. Shneiderman, “Interactive Visualizations for Comparing Two Trees With Structure
and Node Value Changes,” HCIL Tech Report 2011-22, Sep-2011. [Online]. Available: http://hcil2.cs.umd.edu/trs/2011-22/2011-22.pdf.
•
J. A. Guerra-Gomez, A. Buck-Coleman, C. Plaisant, and B. Shneiderman, “TreeVersity: Interactive Visualizations for Comparing Two Trees with
Structure and Node Value Changes,” HCIL Tech Report 2012-04, 2012. [Online]. Available: http://hcil2.cs.umd.edu/trs/2012-04/2012-04.pdf.
•
J. A. Guerra Gómez, M. L. Pack, C. Plaisant, and B. Shneiderman, “Visualizing changes over time in datasets using dynamic hierarchies,” HCIL
Tech Report 2013-06, 2013. [Online]. Available: http://hcil2.cs.umd.edu/trs/2013-06/2013-06.pdf.