Visualization of Shallow Trees with Nodal Attributes using Treemap, Table Lens, and Fisheye Table

1
Visualization of Shallow Treeswith Nodal
Attributesusing Treemap, Table Lens, and
Fisheye Table

• Fred Bernal
• Steve Betten
• Chris Horn
• May 14, 2001

2
Data Hierarchy
3
Spreadsheet View
4
Experiment Variables

• Independent Variable
• Data Visualization Technique
• Treemap
• Table Lens
• Fisheye
• Dependent Variables
• Time to correct completion
• Number of incorrect answers
• Subjective satisfaction
• Subjective ease of use

5
Treemap from HCIL
6
Table Lens from Inxight
7
Fisheye Table from HCIL
8
Experiment Design

• 18 subjects - intermediate computer experience
• 1 x 3 within-subjects design
• 5 minutes of training per treatment
• 4 tasks per treatment
• Answers differed between treatments
• Each subject followed one of 6 possible
  permutations for treatment order
• ABC, ACB, BAC, BCA, CAB, CBA

9
Hypothesis 1 and Result

• Less time to find a maximum value in a sorted
  list than in unsorted 2-d space.
• Question Which unit has the greatest output?
• Fisheye and Table Lens lt Treemap
  

Confirmed
10
Hypothesis 2 and Result

• Less time to find values in a range using filters
  than scanning without using filters.
• Question Out of the units that have an output
  between 20 and 25 (inclusive), how many units
  have at least one problem?
• Treemap lt Table Lens lt Fisheye

Not Confirmed
11
Hypothesis 3 and Result

• Less time to find values in a range by filtering
  and sorting than by filtering alone.
• Question Out of those units that have exactly
  one problem, which unit has the worst (least)
  health?
• Table Lens lt Treemap lt Fisheye Table
  

Not Confirmed
12
Hypothesis 4 and Result

• Less time to find values in a sorted, continuous
  fisheye view than in a sorted, discrete fisheye
  view.
• Which states, if any, contain a county named D?
  Which states, if any, contain a county named G?
• Fisheye Table lt Table Lens lt Treemap

Not Confirmed
13
Hypothesis 5 and Result

• Subjects will be more satisfied with programs
  with greater functionality
• Table Lens gt Treemap gt Fisheye

Confirmed (partially)
14
Hypothesis 6 and Result

• Subjects will find programs with straightforward
  interfaces easier to use
• Treemap Table Lens Fisheye Table

Confirmed
15
Conclusions

• Impact for practitioners
• Filtering
• Suggestions for software developers
• Treemap layout
• Table Lens more interactive filtering
• Fisheye Table reverse sorting, filtering
• Refinement of theory and suggestions for future
  researchers
• General Restrict features for each task
• Fisheye layout Fisheye Table v. Table Lens