Visual Overview Strategies

1
Visual Overview Strategies

• cs5764 Information Visualization
• Chris North

2
Review

• 4 navigation strategies in 1D/2D data?

3
Why Overviews?
Data
Screen
Advantages?
a
a
data
data
a
data
data
detail zoom OD FC trans
4
Advantages of Overviews

• Helps solve the Keyhole Problem
• Spatial map, orientation
• What information is (not) present?
• Adds context info, relationships
• Enables direct access
• Encourages exploration
• HCI metrics
• Improves user performance, learning time,
  satisfaction
• Studies, e.g. BeardWalker, Leung, Plaisant,
  Chimera, North, etc.

5
Visual Overview Design Goals

• Visual take advantage of human visual
  processing
• Information Rich show as much as you can!
  (while maintaining a clean design)
• Interaction Affordances enable quick access to
  details
• E.g. Zooming, OverviewDetail, FocusContext

6
Data Scalability

• Small scalability data easy
• Just show everything
• But, theres always more data
• How much can you show?

7
Overview Strategies for Scalability
Data
Screen

• Screen Reduce visual representation size
• Pack more on the screen
• Data Reduce data quantity
• Use less data to fit screen

8
1. Reduce Visual Representation

• Hammer

Data
Screen
9
E.g. SeeSoft

• 1 line of pixels / line of code

10
Reduce Even More

• Stasko, Information Mural

11
(No Transcript)
12
2. Reduce Data Quantity

• Reduce data quantity to fit screen
• Reduce attributes
• Reduce items
• Reduce value range
• 2 Approaches
• 2.1 Eliminate
• 2.2 Aggregate

13
2.1 Eliminate

• Chainsaw

Data
Screen
14
2.1 Eliminate

• Eliminate items
• Query, Filter, Representative set
• VIDA (Visual Info Density Adjuster) show high
  priority items (video)
• Human-Eye View focused info density
• Eliminate attributes
• Scatterplot selects 2 attributes, ignores rest
• Spotfire use DQ for other attributes
• and Details on demand
• Problem lossy

15
2.2 Aggregate
Combine
d
Screen
Data
16
2.2 Aggregate

• Aggregate items (Group many items into one)
• Which to group together?
• Categorical (SQL group by)
• Spatial (Pad, TableLens, Aggregate Towers)
• Algorithmic (clustering)
• User defined (folders)
• What are attribute values of the groups?
• Math function of group members values (SQL
  group by)
• count, mean, min, max, cluster algorithm
• Abstraction, semantic
• Multiple levels of grouping tree
• Navigation
• Snap-Together Visualization drill down (1M)
• Zooming, Aggregate Towers
• Semantic zooming (Pad, Jazz)

17
(No Transcript)
18
Ward et al. http//davis.wpi.edu/xmdv/movies/
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)
22
2.2 Aggregate

• Aggregate attributes
• Column math grade (hw1 hw2) / 2
• Star Coordinates vector summaps n attributes
  to 2 (x, y)
• Multi-dimensional scalingstatistical technique
  to map n-D to 1,2,3-D usingdistance between
  points

23
Aggregation with Zooming

• Rayson, Aggregate Towers

24
InfoMural vs. TableLens
25
Breakdown Visualization

• Conklin

26
(No Transcript)
27
Summary

• Reduce visual representation Hammer
• Reduce data scale (items,attrs)
• Eliminate Chainsaw
• Aggregate Combine