I590 Interactive Visual Analytics Week 8 | Oct 12, 2016 Multiple - - PowerPoint PPT Presentation

▶
i590 interactive visual analytics
SMART_READER_LITE
LIVE PREVIEW

I590 Interactive Visual Analytics Week 8 | Oct 12, 2016 Multiple - - PowerPoint PPT Presentation

Feb 01, 2024 154 likes •1.92k views

I590 Interactive Visual Analytics Week 8 | Oct 12, 2016 Multiple Views Visualizing Tabular Data Khairi Reda | redak@iu.edu School of Informa5cs & Compu5ng, IUPUI Correction Color mapping for ordered, quantitative data Color mapping

slide-1
SLIDE 1

Khairi Reda | redak@iu.edu School of Informa5cs & Compu5ng, IUPUI

Week 8 | Oct 12, 2016 Multiple Views Visualizing Tabular Data

I590 Interactive Visual Analytics

slide-2
SLIDE 2

Correction…

slide-3
SLIDE 3

Color mapping

for ordered, quantitative data

slide-4
SLIDE 4

Color mapping

Rainbow colormaps should be avoided as a default op5on for ordered data

for ordered, quantitative data

slide-5
SLIDE 5

Color mapping

Rainbow colormaps should be avoided as a default op5on for ordered data A safer, more effec5ve op5on is a colormap that varies in luminance. Ideally luminance and satura3on.

for ordered, quantitative data

slide-6
SLIDE 6

Color mapping

Rainbow colormaps should be avoided as a default op5on for ordered data A safer, more effec5ve op5on is a colormap that varies in luminance. Ideally luminance and satura3on.

hue

for ordered, quantitative data

slide-7
SLIDE 7
  • Important of labeling:
  • Label each chart (axes and units)
  • Show legend when depic5ng mul5ple variables in one chart
  • Provide a descrip5on of the visualiza5on

Project 1 feedback

slide-8
SLIDE 8
  • Important of labeling:
  • Label each chart (axes and units)
  • Show legend when depic5ng mul5ple variables in one chart
  • Provide a descrip5on of the visualiza5on

Project 1 feedback

slide-9
SLIDE 9
  • Important of labeling:
  • Label each chart (axes and units)
  • Show legend when depic5ng mul5ple variables in one chart
  • Provide a descrip5on of the visualiza5on

Project 1 feedback

slide-10
SLIDE 10

Project 1 feedback

  • Make sure there is enough contrast in lightness between

the background and the data.

  • Shape percep5on is based on the lightness channel

(consult chapter 4 of Colin Ware’s book)

slide-11
SLIDE 11

Project 1 feedback

  • Make sure there is enough contrast in lightness between

the background and the data.

  • Shape percep5on is based on the lightness channel

(consult chapter 4 of Colin Ware’s book) weak contrast against background

slide-12
SLIDE 12

Project 1 feedback

  • Avoid using saturated primaries (e.g., pure red or pure

yellow). Instead, use less saturated pastels over white backgrounds

slide-13
SLIDE 13

Project 1 feedback

  • Avoid using saturated primaries (e.g., pure red or pure

yellow). Instead, use less saturated pastels over white backgrounds

slide-14
SLIDE 14

Project 1 feedback

  • Avoid using saturated primaries (e.g., pure red or pure

yellow). Instead, use less saturated pastels over white backgrounds

slide-15
SLIDE 15
  • ScaQerplot is typically used to illustrate two, non-

temporal variables. For 5me, use line chart to illustrate trend

Project 1 feedback

slide-16
SLIDE 16
  • ScaQerplot is typically used to illustrate two, non-

temporal variables. For 5me, use line chart to illustrate trend

Project 1 feedback

slide-17
SLIDE 17

Project 1 feedback

  • ScaQerplot is typically used to illustrate two, non-

temporal variables. For 5me, use line chart to illustrate trend

slide-18
SLIDE 18

Project 1 feedback

  • When showing 5me-varying data, set the axes to the

maximum data range from the onset to avoid “jumping” and improve visual coherence

slide-19
SLIDE 19

Project 1 feedback

  • When showing 5me-varying data, set the axes to the

maximum data range from the onset to avoid “jumping” and improve visual coherence

slide-20
SLIDE 20
slide-21
SLIDE 21

Project 2

Project requirements are somewhat vague, compared to project 1 Just like in the real-world, tasks are never clear in the beginning Understand the data, and think about meaningful analysis tasks that you can facilitate with the visualiza5on

Think about design!

Sketch and evaluate alterna5ve designs before jumping into code Visual encodings have to be appropriate to data types/ tasks

slide-22
SLIDE 22

Project 3

Start thinking about project 3 (out Nov 2)

  • You will have the freedom to define a project, and/or

choose a dataset based on your own interests

  • Can be:
  • Visualiza5on design project
  • User study / evalua5on of an exis5ng technique
  • Experiment inves5ga5ng a fundamental ques5on about

percep5on/cogni5on

  • Team project: max 3 people per team. Have to form

different groups

  • Project proposals: 1 page / team (due Nov 9)
  • Can be same as project 1
  • Proposals will be presented and refined in class
slide-23
SLIDE 23

2 weeks ago

Channels and Marks

slide-24
SLIDE 24

iden3ty channels magnitude channels

Tamara Munzner Via Miriah Meyer

slide-25
SLIDE 25

iden3ty channels magnitude channels

Tamara Munzner Via Miriah Meyer

good for ordered attributes

slide-26
SLIDE 26

iden3ty channels magnitude channels

Tamara Munzner Via Miriah Meyer

good for ordered attributes good for categorical attributes

slide-27
SLIDE 27

How much longer?

Alex Lex

slide-28
SLIDE 28

How much longer?

4x

Alex Lex

slide-29
SLIDE 29

How much larger (area)?

Alex Lex

slide-30
SLIDE 30

How much larger (area)?

5x

Alex Lex

slide-31
SLIDE 31

2 weeks ago

Channels and Marks

slide-32
SLIDE 32

2 weeks ago

Channels and Marks Interac5on

slide-33
SLIDE 33

Types of Interaction

Based on a slide by Alex Lex

Single View

  • Change over 5me
  • Naviga5on
  • Seman5c Zooming
  • Focus+Context
  • Filtering and Querying
slide-34
SLIDE 34

`

Change over Time

Time varying data (animation)

slide-35
SLIDE 35

`

Change over Time

Time varying data (animation)

slide-36
SLIDE 36

Navigation

Pan and Zoom

slide-37
SLIDE 37

Navigation

Pan and Zoom

slide-38
SLIDE 38

Fisheye lenses

slide-39
SLIDE 39

Filtering and Dynamic Query

slide-40
SLIDE 40

Filtering and Dynamic Query

slide-41
SLIDE 41

Filtering and Dynamic Query

slide-42
SLIDE 42

Filtering and Dynamic Query

slide-43
SLIDE 43

One more type of interaction…

Selecting / Highlighting

slide-44
SLIDE 44

One more type of interaction…

Selecting / Highlighting

slide-45
SLIDE 45

One more type of interaction…

Selecting / Highlighting

slide-46
SLIDE 46

Types of Interaction

Based on a slide by Alex Lex

Single View

  • Change over 5me
  • Naviga5on
  • Seman5c Zooming
  • Focus+Context
  • Filtering and Querying
slide-47
SLIDE 47

Types of Interaction

Based on a slide by Alex Lex

Single View

  • Change over 5me
  • Naviga5on
  • Seman5c Zooming
  • Focus+Context
  • Filtering and Querying

Mul3ple Views

  • Selec5on (Details on

Demand)

  • Brushing & Linking
  • Adap5ve

Representa5on

slide-48
SLIDE 48

Types of Interaction

Based on a slide by Alex Lex

Single View

  • Change over 5me
  • Naviga5on
  • Seman5c Zooming
  • Focus+Context
  • Filtering and Querying

Mul3ple Views

  • Selec5on (Details on

Demand)

  • Brushing & Linking
  • Adap5ve

Representa5on

“change over time”

slide-49
SLIDE 49

Types of Interaction

Based on a slide by Alex Lex

Single View

  • Change over 5me
  • Naviga5on
  • Seman5c Zooming
  • Focus+Context
  • Filtering and Querying

Mul3ple Views

  • Selec5on (Details on

Demand)

  • Brushing & Linking
  • Adap5ve

Representa5on

“variety over space” “change over time”

slide-50
SLIDE 50

Views

slide-51
SLIDE 51

Views

One

slide-52
SLIDE 52

Views

One Mul3ple

slide-53
SLIDE 53

Views

One

Eye over memory: use display space instead of working memory

Mul3ple

slide-54
SLIDE 54

One form, multiple views

Par55on data into subsets and distribute among different views Visual Encoding is the same in all views

Nick Elprin, Domino

slide-55
SLIDE 55

One form, multiple views

Par55on data into subsets and distribute among different views Visual Encoding is the same in all views

Nick Elprin, Domino

slide-56
SLIDE 56

One form, multiple views

Par55on data into subsets and distribute among different views Visual Encoding is the same in all views

Nick Elprin, Domino

slide-57
SLIDE 57

One form, multiple views

Par55on data into subsets and distribute among different views Visual Encoding is the same in all views

Small Multiples

Nick Elprin, Domino

slide-58
SLIDE 58

Small-Multiples

Spark Lines

Viz Wiz

slide-59
SLIDE 59

Small-Multiples

Drought, 1898-2012

Mike Bostock

slide-60
SLIDE 60

Small-Multiples

ScaQerplot Matrix

Mike Bostock

x y z w w z y x

Example: dataset with four variables: X, Y, Z, W Par55on aQributes (or variables) and distribute them among different views

slide-61
SLIDE 61

Multi form

Show mul5ple representa5ons of the data Usually the views share the same data Views have different visual encoding (and oien depict different aQributes) Ra3onale: it is difficult to show all aQributes in a single monolithic view. Mul5form views give us freedom to use different visual encodings for different aQributes.

Based on a slide by Miriam Meyer and Alex Lex

slide-62
SLIDE 62

Multi form

MatrixExplorer

Henry 2006 Via Alex Lex

Same data, but different visual encodings (mul5 form)

slide-63
SLIDE 63

Multi form

MizBee

Meyer, 2009

Subsets of the data, but different scales

slide-64
SLIDE 64

View Linking

Views cab be linked implicitly through interac5ons Changes in one view are coordinated to all

  • ther views
slide-65
SLIDE 65

Linked Selection and Highlighting

slide-66
SLIDE 66

Brushing and Linking

Mike Bostock

slide-67
SLIDE 67

Linked Navigation

Overview + Detail

slide-68
SLIDE 68

Linked Navigation

Stacked Zooming

slide-69
SLIDE 69

View Linking

slide-70
SLIDE 70

View Linking

Views can be linked explicitly through visual links Links typically connect the same (or similar) data items in different views

slide-71
SLIDE 71

View Linking

Views can be linked explicitly through visual links Links typically connect the same (or similar) data items in different views Limita3ons: can occlude and lead to visual cluQer, although smart algorithms can route links to minimize side effects

Steinberger et al., 2011 Geymayer et al, 2014

slide-72
SLIDE 72

Details on Demand

Showing addi5onal informa5on with popup views

slide-73
SLIDE 73

TRELLIS

Par33oning variables Par55on the data among mul3ple variables Each view shows a subset of the data Views can be merged or reordered to facilitate paQern percep5on

slide-74
SLIDE 74
slide-75
SLIDE 75
slide-76
SLIDE 76

HiVE

Par55on mul5-aQribute data into a hierarchy Uses treemap as space-filling rectangular layout

Hierarchical Visual Expression

Based on a slide by Miriah Meyer

slide-77
SLIDE 77

Treemap

http://ukdataexplorer.com/co2/

slide-78
SLIDE 78

Treemap

MarketWatch Market Map (defunct, unfortunately)

slide-79
SLIDE 79

Treemap

Disk Inventory X

slide-80
SLIDE 80

HiVE

Partitioning house type neighborhood sale time Property prices in London

Slingsby et al., 2009

Hierarchical Visual Expression

slide-81
SLIDE 81

HiVE

Slingsby et al., 2009

Partitioning neighborhood loc. neighborhood housetype sale time

Hierarchical Visual Expression Property prices in London

slide-82
SLIDE 82

Slingsby et al., 2009

slide-83
SLIDE 83

Layering

Embedding Views in the same space

Combining mul5ple views on top of one another to form a composite view Ra3onale: Supports a larger, more detailed view than using mul5ple views Makes it easier to compare data if mapped to the same axis Tradeoff: Imposes constraints on visual encoding choices and may cause visual cluQer

Based on a slide by Miriah Meyer

slide-84
SLIDE 84

Layering

Embedding Views in the same space

slide-85
SLIDE 85

Layering

Embedding Views in the same space

slide-86
SLIDE 86

Layering

Embedding Views in the same space

NodeTrix, Henry abd Fekete, 2007

slide-87
SLIDE 87

Layering

Embedding Views in the same space

Yost et al., 2007

slide-88
SLIDE 88

Stacking

Mike Bostock

slide-89
SLIDE 89

Stacking

NYTimes

slide-90
SLIDE 90

Critique

NYTimes

http://tinyurl.com/363q6r

slide-91
SLIDE 91

Domino

Graz et al, 2014

Dynamic View creation and linking

slide-92
SLIDE 92
slide-93
SLIDE 93

Highly recommended!

slide-94
SLIDE 94

Data types

slide-95
SLIDE 95

Data types

slide-96
SLIDE 96

Table semantics

slide-97
SLIDE 97

Table semantics

Key aQribute

slide-98
SLIDE 98

Arrange Tables

slide-99
SLIDE 99

Arrange Tables

slide-100
SLIDE 100

Arrange Tables

slide-101
SLIDE 101

iden3ty channels magnitude channels

Tamara Munzner

good for ordered attributes good for categorical attributes

slide-102
SLIDE 102

iden3ty channels magnitude channels

Tamara Munzner

good for ordered attributes good for categorical attributes

slide-103
SLIDE 103

Arrange Tables

slide-104
SLIDE 104

1 aRribute

Arrange Tables

slide-105
SLIDE 105

1 aRribute

Arrange Tables

slide-106
SLIDE 106

2 aRributes 1 aRribute

Arrange Tables

slide-107
SLIDE 107

Arrange Tables

slide-108
SLIDE 108

Arrange Tables

slide-109
SLIDE 109

Arrange Tables

slide-110
SLIDE 110

key attribute

Arrange Tables

slide-111
SLIDE 111

key attribute

Arrange Tables

slide-112
SLIDE 112

key attribute key attribute

Arrange Tables

slide-113
SLIDE 113

Arrange Tables

slide-114
SLIDE 114

Arrange Tables

slide-115
SLIDE 115

Arrange Tables

slide-116
SLIDE 116

Arrange Tables

Mul5ple Keys

Streit & Gehlenborg, PoV, Nature Methods, 2014 Via Alex Lex

slide-117
SLIDE 117

Arrange Tables

Mul5ple Keys

Streit & Gehlenborg, PoV, Nature Methods, 2014 Via Alex Lex

slide-118
SLIDE 118

Arrange Tables

Mul5ple Keys

Streit & Gehlenborg, PoV, Nature Methods, 2014 Via Alex Lex

slide-119
SLIDE 119

Arrange Tables

Mul5ple Keys

Streit & Gehlenborg, PoV, Nature Methods, 2014 Via Alex Lex

slide-120
SLIDE 120

Miriah Meyer

Don’t use line charts for categorical aRributes!

slide-121
SLIDE 121

Miriah Meyer

Don’t use line charts for categorical aRributes!

  • k: “Men are taller than

women (on average)”

slide-122
SLIDE 122

Miriah Meyer

Don’t use line charts for categorical aRributes!

bad: “The more male a person is, the taller he/she is”

  • k: “Men are taller than

women (on average)”

slide-123
SLIDE 123

Miriah Meyer

Don’t use line charts for categorical aRributes!

bad: “The more male a person is, the taller he/she is”

  • k: “Men are taller than

women (on average)”

slide-124
SLIDE 124

Miriah Meyer

Don’t use line charts for categorical aRributes!

bad: “The more male a person is, the taller he/she is”

  • k: “Men are taller than

women (on average)”

  • k: “Twelve year olds are

taller than ten years old”

slide-125
SLIDE 125

Miriah Meyer

Don’t use line charts for categorical aRributes!

bad: “The more male a person is, the taller he/she is”

  • k: “Men are taller than

women (on average)”

  • k: “Twelve year olds are

taller than ten years old”

  • k: “Height increases

with age”

slide-126
SLIDE 126

Arrange Tables

Table as a heatmap

1 2 5 4 5 1 5 6 1 2 2 1 3 1 4 1 2 1

slide-127
SLIDE 127

Arrange Tables

Table as a heatmap

1 2 5 4 5 1 5 6 1 2 2 1 3 1 4 1 2 1

slide-128
SLIDE 128

Arrange Tables

Table as a heatmap

slide-129
SLIDE 129

Arrange Tables

Table as a heatmap

slide-130
SLIDE 130

Arrange Tables

Table as a heatmap

slide-131
SLIDE 131

Arrange Tables

Table as a heatmap Order is important: Clustering is oien used with heatmaps

slide-132
SLIDE 132

Arrange Tables

Align using mul5ple keys

Gratzl et et. 2013

slide-133
SLIDE 133

Arrange Tables

slide-134
SLIDE 134

Arrange Tables

slide-135
SLIDE 135

Wilkinson et al., 2005 Via Miriah Meyer

slide-136
SLIDE 136

Wilkinson et al., 2005 Via Miriah Meyer

slide-137
SLIDE 137

nine characteristics of Abalone (sea snails)

Wilkinson et al., 2005 Via Miriah Meyer

slide-138
SLIDE 138

Wilkinson et al., 2005 Via Miriah Meyer

slide-139
SLIDE 139

Wilkinson et al., 2005 Via Miriah Meyer

slide-140
SLIDE 140

Example by Miriah Meyer

Parallel Coordinates

slide-141
SLIDE 141

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-142
SLIDE 142

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-143
SLIDE 143

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-144
SLIDE 144

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-145
SLIDE 145

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-146
SLIDE 146

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-147
SLIDE 147

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-148
SLIDE 148

Example by Miriah Meyer

V1 V2 V3 V4 V5

2 4 6 8 10

Parallel Coordinates

slide-149
SLIDE 149

Wegman 1990 Via Miriah Meyer

posi3ve correla3on straight lines nega3ve correla3on all lines cross at a single point

Parallel Coordinates

slide-150
SLIDE 150

ProtoVis Via Miriah Meyer

Parallel Coordinates

slide-151
SLIDE 151

Fua 1999 Via Miriah Meyer

Do you see any correlation?

slide-152
SLIDE 152

Fua 1999 Via Miriah Meyer

Do you see any correlation?

Correla3ons only visible between neighboring axis pairs:

slide-153
SLIDE 153

Fua 1999 Via Miriah Meyer

Do you see any correlation?

Correla3ons only visible between neighboring axis pairs:

slide-154
SLIDE 154

Fua 1999 Via Miriah Meyer

Do you see any correlation?

Correla3ons only visible between neighboring axis pairs: axis order maQers

slide-155
SLIDE 155

Fua 1999 Via Miriah Meyer

Do you see any correlation?

Correla3ons only visible between neighboring axis pairs: axis order maQers

slide-156
SLIDE 156

Fua 1999 Via Miriah Meyer

Do you see any correlation?

Correla3ons only visible between neighboring axis pairs: axis order maQers allow user to reorder axis

slide-157
SLIDE 157

Hierarchical Parallel Coordinates

Fua 1999

slide-158
SLIDE 158

Hierarchical Parallel Coordinates

Fua 1999

slide-159
SLIDE 159

Hierarchical Parallel Coordinates

Fua 1999

slide-160
SLIDE 160

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster:

slide-161
SLIDE 161

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster:

slide-162
SLIDE 162

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster: mean: opaque line

slide-163
SLIDE 163

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster: mean: opaque line

slide-164
SLIDE 164

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster: mean: opaque line min/max: illustrated by band width with decreasing opacity from mean

slide-165
SLIDE 165

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster: mean: opaque line min/max: illustrated by band width with decreasing opacity from mean

slide-166
SLIDE 166

Hierarchical Parallel Coordinates

Fua 1999

Instead of showing all points, show a band represen3ng a cluster: mean: opaque line min/max: illustrated by band width with decreasing opacity from mean cluster

slide-167
SLIDE 167

Hierarchical Parallel Coordinates

Fua 1999

Cluster: lines that share similar shapes. Interac5vely varying the similarity threshold allows us to “unpack” clusters

slide-168
SLIDE 168
slide-169
SLIDE 169
slide-170
SLIDE 170

Radial Layout

Donut Charts

slide-171
SLIDE 171

Radial Layout

Star Plot

Similar to parallel coordinates, but axes radiate from a common origin

Via Alex Lex

slide-172
SLIDE 172

Radial Layout

Star Plot

Similar to parallel coordinates, but axes radiate from a common origin

Scotch Whiskies

Via Alex Lex

slide-173
SLIDE 173

Arrange Tables

slide-174
SLIDE 174

Arrange Tables

slide-175
SLIDE 175

Next week

Spa3al Data Chapter 8 Networks Chapter 9