Table Lens

1
Table Lens

• Introduction to the Table Lens concept
• Table Lens Implementation
• Projected Usage Scenarios
• Usage Comparison with Splus
• Critical Analysis

2
Interfacing with Tables

• Tables as a common representation
• Regularized content
• Instance vs. Value Layout
• Large raw tables are uninterpretable

3
Table Views

• Two common modes of interface
• Focus User examines as many fields as will fit
  on her screen (660 cells)
• Sacrifices relationships for detail
• Context A generated representation of broad
  dataset it examined
• High-level interpretation with no specifics

4
Table Lens

• Lenses are devices which focus attention on part
  of a large context
• Table Lens allows both Focus and Context views
  simultaneously
• Tables are regular
• Deformations are also regular
• Lensing creates categories of detail

5
Categories of detail

• Central areas of lens have highest levels of
  detail
• Row and Column focal have less detail
• Non-focal areas have sharply reduced detail but
  are still present

6
Degree of Interest

• Detail categorization and Visualization based on
  Degree of Interest calculation
• DOI Translates to cell size along 2 independent
  axes
• Binary correspondence in each dimension

7
Table Lens Benefits

• Table Lens can display both focus and context
• Much more data can be displayed at once
• 30 100 times basic spreadsheet
• Allows simultaneous view of
• Variable value distribution shape
• Inter-variable correlation
• Specific instance values
• Outlier identification

8
(No Transcript)
9
Table Lens Implementation

• Interactive manipulation of focus Atomic
  operations
• Zoom Enlargement of focal area
• Ajust Expansion of focal contents
• Slide Positioning of focal area
• Composite manipulation
• Adjust-zoom Adds items to focus while expanding
  focal area

10
Multiple Foci

• Multiple focal areas are supported
• Important use modifications
• Adjust corrupts display
• Zoom required to be global

11
(No Transcript)
12
Graphical Cell Representation

• Presentation factors
• Value
• Value Type
• DOI (Region) Type
• Cell Size
• User Choices
• Spotlighting

13
Other Features

• Ascending/Descending Sorting
• Spotlighting
• Formula compilation
• Median, Quarter, Extents Selection

14
Table Lens facilitates

• Correlation of variable value curves
• Outlier identification/interrogation
• Variable nesting identification
• Ease of use (Simple!)

15
Usage Comparison

• Exploratory Data Analysis (EDA)
• Sensemaking
• Activities in which external representations
  are interpreted into semantic content and
  represented in some other manner

16
EDA Tasks

• Batch Assessment
• Determining structure of information and its
  irregularities
• Variable modeling
• Finding formulaic expression for variable values

17
Learning Loop

• Steps
• Search for representation of regularities
• Encoding information into representation
• Altering representation to accommodate outliers
• Use of representation for discovery

18
Table Lens vs Splus

• Estimating utility of application approach
• Required time to perform tasks
• Benchmark times
• Empirical times
• Qualitative Considerations
• Ease of use
• Complexity vs Return

19
Correlation Table Lens
20
Correlation Splus
21
Time-cost for important properties of all
variables

• Table Lens superior for iterative analysis
• Splus faster for random access

22
Time-cost for related variables

• Table Lens superior when several clusters can be
  grouped and eliminated early
• Splus more effective when broad dataset must be
  analyzed

23
Learning costs

• Table Lens performs within significant margins as
  well as Splus
• Table Lens is much simpler than Splus