Whats the Difference?: Evaluating Variants of Multi-Series Bar - - PowerPoint PPT Presentation

▶
what s the difference
SMART_READER_LITE
LIVE PREVIEW

Whats the Difference?: Evaluating Variants of Multi-Series Bar - - PowerPoint PPT Presentation

Mar 09, 2024 475 likes •1.14k views

Whats the Difference?: Evaluating Variants of Multi-Series Bar Charts for Visual Comparison Tasks Arjun Srinivasan Matthew Brehmer Bongshin Lee Steven M. Drucker Whats changed? Year over Year sales Production in Region 1 vs.

slide-1
SLIDE 1

What’s the Difference?:

Evaluating Variants of Multi-Series Bar Charts for Visual Comparison Tasks

Arjun Srinivasan Steven M. Drucker Matthew Brehmer Bongshin Lee

slide-2
SLIDE 2
slide-3
SLIDE 3
slide-4
SLIDE 4

“What’s changed?”

Year over Year sales Production in Region 1 vs. Production in Region 2 …

slide-5
SLIDE 5
slide-6
SLIDE 6
slide-7
SLIDE 7
slide-8
SLIDE 8
slide-9
SLIDE 9

How can we facilitate visual comparison in multi-series bar charts?

slide-10
SLIDE 10

+

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Constant Varying

Interviews & Survey Exploration of design alternatives Evaluation Applications & Extensions

slide-11
SLIDE 11

+

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Constant Varying

Interviews & Survey Exploration of design alternatives Evaluation Applications

slide-12
SLIDE 12
slide-13
SLIDE 13

Bar charts were the most common visualizations

Ordinal x Quantitative (e.g., monthly sales) [constant] Nominal x Quantitative (e.g., individual employee sales) [varying]

slide-14
SLIDE 14

Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

Design considerations:

  • Show raw values

Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

slide-15
SLIDE 15

Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4

Design considerations:

  • Show raw values
  • Do not occupy additional space
slide-16
SLIDE 16

P1 P2 P3 P4 P1 P2 P3 P4 Year1 Year2 1 2 3 4 P1 P2 P3 P4

Design considerations:

  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
slide-17
SLIDE 17

Design considerations:

  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for varying vs. constant data conditions
  • Make it easier to measure differences
slide-18
SLIDE 18
slide-19
SLIDE 19
slide-20
SLIDE 20
slide-21
SLIDE 21
slide-22
SLIDE 22
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
slide-23
SLIDE 23
slide-24
SLIDE 24
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for data conditions
  • Make it easier to measure difference
slide-25
SLIDE 25
slide-26
SLIDE 26
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for data conditions
  • Make it easier to measure difference
slide-27
SLIDE 27
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for variations in data
  • Make it easier to measure difference
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for variations in data
  • Make it easier to measure difference
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for variations in data
  • Make it easier to measure difference

???

slide-28
SLIDE 28

+

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Constant Varying

Interviews & Survey Exploration of design alternatives Evaluation Applications

slide-29
SLIDE 29

Juxtaposition Explicit Encoding Superimposition

M Gleicher et al. 2011

slide-30
SLIDE 30

Juxtaposition Explicit Encoding Superimposition

slide-31
SLIDE 31

BarTender

Juxtaposition Explicit Encoding Superimposition

slide-32
SLIDE 32

BarTender

Juxtaposition Explicit Encoding Superimposition

slide-33
SLIDE 33

BarTender

Juxtaposition Explicit Encoding Superimposition

slide-34
SLIDE 34

BarTender

Juxtaposition Explicit Encoding Superimposition

slide-35
SLIDE 35

BarTender

Juxtaposition Explicit Encoding Superimposition + +

slide-36
SLIDE 36

BarTender

Juxtaposition Explicit Encoding Superimposition + +

slide-37
SLIDE 37

BarTender

Juxtaposition Explicit Encoding Superimposition +

slide-38
SLIDE 38
slide-39
SLIDE 39

Grouped Bar Chart Difference Chart

slide-40
SLIDE 40

Grouped Bar Chart w/ difference overlay Single Bar Chart w/ difference overlay

slide-41
SLIDE 41
  • Show raw values
  • Do not occupy additional space
  • Maintain visualization type
  • Account for variations in data
  • Make it easier to measure difference

Grouped Bar Chart w/ difference overlay Single Bar Chart w/ difference overlay

slide-42
SLIDE 42

+

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Constant Varying

Interviews & Survey Exploration of design alternatives Evaluation Applications

slide-43
SLIDE 43

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Constant Varying

74 Participants 6 Tasks 2 Data Conditions

slide-44
SLIDE 44

Tasks

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories

slide-45
SLIDE 45

Value interpretation Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories

Tasks

slide-46
SLIDE 46

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories

Tasks

slide-47
SLIDE 47

Difference-based Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories

Tasks

slide-48
SLIDE 48

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories

Tasks

slide-49
SLIDE 49

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Varying data conditions

Tasks

slide-50
SLIDE 50

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories

Tasks

slide-51
SLIDE 51

Design Introduction Data Type 1 Introduction Training Phase Testing Phase Data Type 2 Introduction Training Phase Testing Phase

slide-52
SLIDE 52

Design Introduction Data Type 1 Introduction Training Phase Testing Phase Data Type 2 Introduction Training Phase Testing Phase

slide-53
SLIDE 53

Design Introduction Data Type 1 Introduction Training Phase Testing Phase Data Type 2 Introduction Training Phase Testing Phase

slide-54
SLIDE 54

Design Introduction Data Type 1 Introduction Training Phase Testing Phase Data Type 2 Introduction Training Phase Testing Phase

slide-55
SLIDE 55

Design Introduction Data Type 1 Introduction Training Phase Testing Phase Data Type 2 Introduction Training Phase Testing Phase

slide-56
SLIDE 56

Design Introduction Data Type 1 Introduction Training Phase Testing Phase Data Type 2 Introduction Training Phase Testing Phase

4 visualizations 87 trials (28 training, 56 testing, 3 guess checking)

(time) (error) (subjective preferences)

slide-57
SLIDE 57

Results

vs

  • Comparable for value

interpretation and varying data condition tasks.

  • Performance with hybrid designs

was better for difference-based tasks.

slide-58
SLIDE 58

Results

vs

  • Comparable for

difference-based tasks.

slide-59
SLIDE 59

Results

Which of the four chart designs did you prefer most?

slide-60
SLIDE 60

Results

Which of the four chart designs did you prefer least?

slide-61
SLIDE 61

+

Extremes- Target Extremes- Source

  • Max. Change

Value Difference Old Categories New Categories Constant Varying

Interviews & Survey Exploration of design alternatives Evaluation Applications & Extensions

slide-62
SLIDE 62

Revealing changes in narrative visualizations

slide-63
SLIDE 63

Complementing overlays with annotations for missing values

slide-64
SLIDE 64

Summary

  • Visual comparison is an important task in dashboards
  • Hybrid visualizations combining design strategies afford more tasks while performing

comparably on individual tasks.

slide-65
SLIDE 65

Summary

  • Visual comparison is an important task in dashboards
  • Hybrid visualizations combining design strategies afford more tasks while performing

comparably on individual tasks.

vs vs

slide-66
SLIDE 66

Arjun Srinivasan

http://arjun010.github.io/

Matthew Brehmer Steven M. Drucker Bongshin Lee

Thank You