Trees and Cushions

1
Trees and Cushions

• Jack van Wijk
• Eindhoven University of Technology
• Treemap Workshop, May 31, 2001
• HCIL, University of Maryland

2
InfoVis at Eindhoven

• Started 1998
• Focus
• Trees and graphs
• Large data sets
• Use of computer graphics knowledge (textures,
  geometry, shading) to generate more effective
  visualizations

3
Trees (T) and Cushions (C)

• TC Cushion treemaps (InfoVis99)
• TC Squarified treemaps (Vissym00)
• C Voronoi diagrams (Vissym01)
• C Enridged contour maps (Vis01)
• T Botanical vis (InfoVis01)
• What next?

4
Cushion TreemapsVisualization of Hierarchical
Information

• Jarke J. van Wijk
• Huub van de Wetering
• Eindhoven University of Technology
• IEEE InfoVis99

5
Insight in structure of large trees

• Why is my disk full?
• What is our product portfolio?
• How is this university organized?
• Fuzzy questions Visualization needed

6
Treemap (Shneiderman, 1992)
E1
C3
G2
A16
H4
F2
B3
C3
D10
I4
E1
F2
G2
H4
I4
Alternating directions, area represents size
7
1400 files
8
3060 employees
9
How to emphasize structure?

• Color?
• Linewidth?
• Nesting?
• Shading?
• Use shaded geometric model!

10
Ridges for more insight
Binary tree
11
Ridge rotated ridge cushion
12
Standard treemap
13
Cushion Treemap
14
H 0.75
level
15
H 0.50
level
16
Demo
www.win.tue.nl/sequoiaview May 21 2001 45,000
downloads
17
Squarified Treemaps

• Mark Bruls
• Kees Huizing
• Jarke J. van Wijk
• Eindhoven University of Technology
• Vissym00, Amsterdam

18
Thin rectangles

• (small leaves high in hierarchy
• e.g., .cshrc)

• hard to compare sizes
• hard to point at
• waste of pixels
• inaccurate size

19
How to avoid thin rectangles?

• drop the single direction layout

• (emphasize structure by other means)

20
Squarification algorithm

• 1. Start placing recs in one row
• 2. stop when aspect ratio stops getting better
• 3. repeat with remaining area and recs

Recursive per level (just like standard treemap
algorithm)
21
Squarification algorithm
6
6
6
6
4
4
3
4/1
aspect ratio 8/3
3/2
2
2
1
2
3
4
9/2
49/27
9/4
25/9
22
Result of squarification
directory
23
Squarified organization
24
Shaded frames for structure
25
Frames for structure

• no maze running for the viewer
• depth in structure as frame height
• remote cousins are visibly separated by indent

26
Organization
27
Directory structure
28
Visualization of Generalized Voronoi Diagrams

• Alex Telea, Jarke van Wijk
• Vissym01, Ascona

29
Cushions

• Cushions help to understand hierarchical spatial
  tesselations of the plane
• How about cushions to visualize Generalized
  Voronoi Diagrams?

30
Generalized Voronoi diagrams
N 1
N 2
Polygon area where N seeds are closest
31
Cushions and bevels
32
Cushions, bevels, color
33
N 3 Cushions, bevels, color
34
Generalized Voronoi Diagrams

• Many other types (different distance measures)
• Applications

35
Enridged Contour Maps

• Van Wijk Telea, Vis01, San Diego
• Given Height field f(x,y)
• Required
• Qualitative (where are the ridges) and
• Quantitative (how high is this peak) info

36
Standard visualizations
37
Enridged height field ...
height(f(x, y))
linear mapping
non-linear mapping
f(x, y)
38
Height field
39
Soft, convex ridges
40
Strong, convex ridges
41
Soft,concave ridges
42
Climate (January)
Color Temperature Height Precipitation
43
Climate (July)
Color Temperature Height Precipitation
44
Dense contours (equid. in space)
45
With ridges...
46
Hierarchical ridges
47
Back to TreesBotanical Visualization of Huge
Hierarchies

• Ernst Kleiberg, Huub van de Wetering, Jarke van
  Wijk
• InfoVis01, San Diego

48
Idea

• Botanical trees are easy to understand, yet
  contain a lot of branches and leaves
• Can we use ideas from botanical modeling for
  InfoVis?

49
Strand model (Holton, 1994)

• Mimics vascular system
• Each leaf is connected to one strand
• Branch bundle of strands
• Rules define when a branch is split
• First try
• Each directory is a branch
• Each file is a leaf

50
Naive result...
51
Three problems

• Continuing branches are hard to see
• Long, thin branches emerge
• Leaves are messy

52
Smoothed continuing branches
53
Contract long branches
54
Files Phi-balls (Lintermann,99)
Many small files
One big file
55
(No Transcript)
56
(No Transcript)
57
(No Transcript)
58
(No Transcript)
59
Botanical modeling

• Compact, natural visualization
• Phyllotaxis magic!
• Many treasures to be discovered
• Usability?
• Botanical treemaps?

60
My treemap to-do list

• Non-rectangular shapes/subdivisions
• circles, polygons?
• Multivariate data
• color, texture?
• Applications
• genealogy, data mining?
• Evaluation