marks + channelsdap.vsb.cz/cass2017/img/3_thu_0830_vis_encoding... · 2017. 9. 7. · ©...

© Munzner/Möller

Marks + Channels

2017 Czech-Austrian summer school Thomas Torsney-Weir

© Munzner/Möller

Overview• Marks + channels • Channel effectiveness

– Accuracy – Discriminability – Separability – Popout

• Channel characteristics – Spatial position – Colour – Size – Tilt (angle) – Shape (glyph) – Stipple (texture) – Curvature – Motion 2

© Munzner/Möller 3

Readings• Munzner, “Visualization Analysis and Design”:

– Chapter 5 (Marks and Channels) • Colin Ware:

– Chapter 4 (Color) – Chapter 5 (Visual Attention and Information that

Pops Out) • The Visualization Handbook:

– Chapter 1 (Overview of Visualization) • Additional (background) reading

– J. Mackinlay: Automating the design of graphical presentations of relational information. ACM ToG, 5(2), 110-141, 1986

Marks + Channels

Rectangle

Width

Height

Text x-position

y-position

color

Line

© Munzner/Möller

Marks + Channels• Mark: basic graphical element / geometric primitive:

– point (0D) – line (1D) – area (2D) – volume (3D)

• Channel: control appearance (of a mark) – position – size – shape – orientation – hue, saturation, lightness – etc.

7

Height: bin size

x-position: bin name

© Munzner/Möller

Visual Language is a Sign System

• Image perceived as a set of signs • Sender encodes information in signs • Receiver decodes information from

signs

• Jacques Bertin – French cartographer [1918-2010] – Semiology of Graphics [1967] – Theoretical principles for visual encodings

Semiology of Graphics [J. Bertin, 83]

13

Image

Visual language is a sign system

Images perceived as a set of signs

Sender encodes information in signs

Receiver decodes information from signs

Semiology of Graphics, 1983

Jacques Bertin

9

© Munzner/Möller

According to Bertin ...

Position

Size

(Grey)Value

Texture

Color

Orientation

ShapeSemiology of Graphics [J. Bertin, 67]

Points Lines AreasMarks

Cha

nnels

10

© Munzner/Möller

Stolte / Hanrahan

“Polaris: A System for Query, Analysis and Visualization of Multi-dimensional Relational Databases”, Chris Stolte and Pat Hanrahan

11

http://graphics.stanford.edu/papers/polaris/

© Munzner/Möller

Progression

12

© Munzner/Möller

Progression

13

© Munzner/Möller

Progression

14

© Munzner/Möller

Progression

15

© Munzner/Möller

Channel types: What/where

16

Where/how much

What

© Munzner/Möller

What vs. How Much channels• What: categorical

– shape – spatial region – colour (hue)

• How Much: ordered (ordinal, quantitative) – length (1D) – area (2D) – volume (3D) – tilt – position – colour (lightness) 17

© Munzner/Möller

Mark types• tables: item = point • network: node+link • link types:

– connection: relationship btw. two nodes – containment: hierarchy

18

Marks as Items/Nodes

Marks as Links

Points Lines Areas

Containment Connection

© Munzner/Möller

Expressiveness + Effectiveness

• expressiveness principle: – visual encoding should express all of, and

only, the information in the dataset attributes

– lie factor

19

© Munzner/Möller

Expressiveness + Effectiveness

• effectiveness principle: – importance of the attribute should match

the salience of the channel – data-ink ratio

20


Effectiveness of Mappings

• Effectiveness according to neurophysiology

• Cells in Visual Areas 1 and 2 differentially tuned to each of the following properties: – Orientation and size (with luminance) – Color (two types of signal) – Stereoscopic depth – Motion

© Munzner/Möller

Effectiveness -- Accuracy

• perceptual judgement vs. stimulus

• Weber’s law:  S = In

22

Magnitude Channels: Ordered Attributes Identity Channels: Categorical Attributes

Spatial region

Color hue

Motion

Shape

Position on common scale

Position on unaligned scale

Length (1D size)

Tilt/angle

Area (2D size)

Depth (3D position)

Color luminance

Color saturation

Curvature

Volume (3D size)

Channels: Expressiveness Types and Effectiveness Ranks

© Munzner/Möller

Effectiveness -- Discriminability• how many  

colors can  I tell apart?

• how many  levels of  grey etc.

• Ex: line width

24

© Munzner/Möller

Effectiveness -- Separability

• separable vs. integral channels

25


According to Ware ...• Integral display dimensions

– Two or more attributes perceived holistically

• Separable dimensions – Separate judgments about each

graphical dimension • Simplistic classification, with a large  

number of exceptions and asymmetries

More integral coding pairs

More separable coding pairs

[C. W

are,

Info

rmat

ion

Visu

aliza

tion]

© Munzner/Möller

Popout - Preattentive processing

• parallel (visual processing)

27

© Munzner/Möller



28

© Munzner/Möller



29

© Munzner/Möller



30

© Munzner/Möller



31

© Munzner/Möller

Overview• Marks + channels • Channel effectiveness • Channel characteristics

– Spatial position – Color – Size – Tilt (angle) – Shape (glyph) – Stipple (texture) – Curvature – Motion 32

© Munzner/Möller

Channels

• Spatial position: most effective for all data types (remember the power of the plane)

• Size: ‘how much’, interacts with others • Shape/Glyph: ‘what channel’ • Stipple/texture: less popular today • Curvature • Motion: large popout effect

33

© Munzner/Möller

Spatial position

34

2.05D

© Munzner/Möller

Colour

35

Thanks to Moritz Wustinger

Thanks to Moritz WustingerSmiley based on http://upload.wikimedia.org/wikipedia/commons/b/bd/A_Smiley.jpg

http://upload.wikimedia.org/wikipedia/commons/b/bd/A_Smiley.jpg

Color deficiency

© Munzner/Möller

Model “Color blindness” • Flaw in opponent processing

– Red-green common (deuteranope, protanope) – Blue-yellow possible (tritanope -- most common) – Luminance channel almost “normal”

• 8% of all men, 0.5% of all women • Effect is 2D color vision model

– Flatten color space – Can be simulated (Brettel et. al.) – http://colorfilter.wickline.org – http://www.colblindor.com/coblis-color-

blindness-simulator/41

Source: M. Stone

http://colorfilter.wickline.orghttp://www.colblindor.com/coblis-color-blindness-simulator/http://www.colblindor.com/coblis-color-blindness-simulator/

© Munzner/Möller

Color Blindness

DeuteranopeProtanope TritanopeNo M cones No L cones No S cones

Red / green deficiencies

Blue / Yellow deficiency

42

Source: M. Stone

© Munzner/Möller

Color-Blindness

Normal Protanope Deuteranope Lightness43

Source: M. Stone

Using Color

Categorical Data

© Munzner/Möller

Categorical Data

• Limited distinguishability (8-14) – Best with Hue – Best choices from Ware:

Source: Ware, “Information Visualization”

© Munzner/Möller

Maximum Hue Separation

47

Source: H.P. Pfister

© Munzner/Möller

Analogous, yet distinctSource: H.P. Pfister

© Munzner/Möller

Primary Colors

• Primary color terms are remarkably consistent across cultures [Berlin & Kay, 69]

Source: Ware, “Information Visualization”

49

© Munzner/Möller

Take-home message

• Only a small number of colors can be used effectively as categorical labels

• Keep the number of colors for categorical data to less than eight, and use quiet medium grey backgrounds

50


© Munzner/Möller

Small Areas

Tableau Software

51

© Munzner/Möller Tufte, VDQI (Vol. 1), p. 77

Large Areas

52

© Munzner/Möller

Large Areas

53Tufte, VDQI (Vol. 1), p. 77

© Munzner/Möller

Large Areas

Cleveland & McGill, “A Color-Caused Optical Illusion on a Statistical Graph”, 1983

1 = Red is bigger 2 = Green is bigger 3 = Both look the same

54

http://graphics.stanford.edu/papers/polaris/http://graphics.stanford.edu/papers/polaris/

© Munzner/Möller

Color Size Illusion

55Cleveland & McGill, “A Color-Caused Optical Illusion on a Statistical Graph”, 1983

http://graphics.stanford.edu/papers/polaris/http://graphics.stanford.edu/papers/polaris/

© Munzner/Möller

Maps

56

Source: Ware, “Visual Thinking for Design”

© Munzner/Möller

Take-home message

• Color in small regions is difficult to perceive, and bright colors in large areas appear bigger

• Use bright, saturated colors for small regions, and use low saturation pastel colors for large regions and backgrounds

57


Ordinal

© Munzner/Möller

Order These Colors

59

Source: J. Stasko

© Munzner/Möller

Order These Colors

60

Source: J. Stasko

http://colorbrewer2.org/

© Munzner/Möller

Brewer Scales

Cynthia Brewer, Color Use Guidelines for Data Representation

Nominal

Ordinal

62


© Munzner/Möller

SequentialSource: H.P. Pfister

© Munzner/Möller

Take-home message

• Lightness and saturation are effective for ordinal data because they have an implicit perceptual ordering

• Show ordinal data with a discrete set of color values that change in lightness or saturation

64


Quantitative

© Munzner/Möller

Rainbow Colormap

Rogowitz and Treinish, Why should engineers and scientists be worried about color?

66

http://graphics.stanford.edu/papers/polaris/http://graphics.stanford.edu/papers/polaris/http://graphics.stanford.edu/papers/polaris/

© Munzner/Möller

Rainbow Colormap

67

Rogowitz and Treinish, Why should engineers and scientists be worried about color?

http://graphics.stanford.edu/papers/polaris/http://graphics.stanford.edu/papers/polaris/http://graphics.stanford.edu/papers/polaris/

© Munzner/Möller

Rainbow Colormap

• Hue is used to show ordinal data

• Not perceptually linear: Equal steps in the continuous range are not perceived as equal steps

• Not good for colorblind people

68

© Munzner/Möller

Rainbow colour map

• Learned order • Visually segmented

– Solution — isoluminant rainbow – Solution — discretize colormap

69

© Munzner/Möller

Color Segmentation

C. Ware, “Visual Thinking for Design”

70


© Munzner/Möller

ColormapsDensity Map

Lightness scale

Lightness scale

with hue and

chroma variation

Hue scale with

lightness variation

Density Map

Lightness scale

Lightness scale

with hue and

chroma variation

Hue scale with

lightness variation

Density Map

Lightness scale

Lightness scale

with hue and

chroma variation

Hue scale with

lightness variation

After slide from M. Stone 71


© Munzner/Möller

Take-home message

• Quantitative data can be shown with a discrete or continuous colormap

• Use colormaps with a limited hue palette and redundantly vary lightness and saturation, and use discrete colormaps for accuracy

72

Color Aesthetics

© Munzner/Möller

DANGER!

zInappropriate use of colour can be disasterous to the application

74

© Munzner/Möller

Why Should We Care?

• Poorly designed color is confusing – Creates visual clutter – Misdirects attention

• Poor design devalues the information – Visual sophistication – Evolution of document and web design

• Don Norman: “Attractive things work better”

75

Source: M. Stone

© Munzner/Möller

Overview• Marks + channels • Channel effectiveness • Channel characteristics

– Spatial position – Color

• Other channels: – Size – Tilt (angle) – Shape (glyph) – Stipple (texture) – Curvature – Motion 76

Magnitude Channels: Ordered Attributes Identity Channels: Categorical Attributes

Spatial region

Color hue

Motion

Shape

Position on common scale

Position on unaligned scale

Length (1D size)

Tilt/angle

Area (2D size)

Depth (3D position)

Color luminance

Color saturation

Curvature

Volume (3D size)

Channels: Expressiveness Types and Effectiveness Ranks

© Munzner/Möller

Relative vs absolute judgement

• Weber’s law says that everything is relative, i.e. the “intensity” depends on the background signal

78

© Munzner/Möller


81https://thenextweb.com/dd/2015/05/15/7-most-common-data-visualization-mistakes/#.tnw_LZw9olbK

Unframed Framed

83http://de.wikipedia.org/wiki/Optische_Täuschung

http://de.wikipedia.org/wiki/Optische_T%C3%A4uschung

In conclusion• Visualizations are broken down into marks

(elements) and channels (parameters) • Data is linked to these channels • Alot of care in choosing which and how

many channels to use

84


Readings

• Munzner, “Visualization Analysis and Design”: – Chapter 7 (Encode: Arrange in Space) – Chapter 8 (Encode: Spatial Layouts and Link

Marks) – Chapter 9 (Encode: Color and Other

Channels)

© Munzner/Möller

Statistical Graphics

• heavy focus on spatial position for visual encoding

• long history for paper-based views of data, see http://www.datavis.ca/milestones/

• many ways to make interactive • many ways to refine / improve / combine

87

http://www.datavis.ca/milestones/

© Munzner/Möller

1D keys

• categorical: bar charts – aligned / ordered

• quantitative/ ordered: dot plot / line chart

88

© Munzner/Möller

Line charts

• invented by William Playfair (1759-1823) – also invented bar charts, pie charts, etc.

89

http://www.math.yorku.ca/SCS/Gallery/images/playfair-wheat1.gif

http://www.math.yorku.ca/SCS/Gallery/images/playfair-wheat1.gif

© Munzner/Möller

Bar vs. line charts

• line implies trend, not appropriate for categorical data

90Zacks and Tversky. Bars and Lines: A Study of Graphic Communication. Memory and Cognition 27(6):1073-1079, 1999.

© Munzner/Möller

2D keys - scatterplots

• encode two input variables with spatial position – show positive/

negative/ no correlation between variables

– show clusters: clumpiness/density, shape, overlap

91

http://upload.wikimedia.org/wikipedia/commons/0/0f/Oldfaithful3.png

http://upload.wikimedia.org/wikipedia/commons/0/0f/Oldfaithful3.png

© Munzner/Möller

Scagnostics

92

Figu

re3:

Scaled

graph-th

eoreticm

easures

(blu

e=low

,red

=high

)for

elevenscatter

pattern

s

pointsetstend

toincrease

computation

time

becauseof

theneed

forrebinning.Datasets

with

compact,nonuniform

distributionsare

considerablyfasterto

compute.

5E

XA

MPL

ES

Figure5

shows

ascagnostics

SPLOM

fortheA

balonedataset.W

ehave

highlighteda

pointinthe

SPLOM

thatrepresentsoneunusual

scatterplot.Thelinked

plotisshown

inthe

upperrightofthefigure.

Fromw

hatwe

seein

theSPLO

M,w

ecan

characterizethis

scatter-plotas

relativelyhigh

inoutliers,skew

edin

edgelengths,clum

py,nonconvex,striated,and

stringy.Figure

6show

sa

scagnosticsSPLO

Mof

17variables

froma

datasetbased

onstatistics

forselected

countriescom

piledby

theW

orldH

ealthO

rganizationand

theU

nitedN

ations[43].

We

havehighlighted

apoint

inthe

graph-theoreticSPLO

Mto

givean

ex-am

pleof

thetype

ofanom

aliesthatscagnostics

canuncover.

Thered

pointisclearly

anoutlierin

mostofthe

panels.Thescatterplot

Figu

re4:

Bip

lotof

measu

resan

dscatters

correspondingto

thispointis

shown

inthe

upperrightof

thefig-

ure.W

ew

eresurprised

tofind

thatwe

hadincluded

two

versionsof

acategoricalvariable

measuring

thetype

ofgovernm

ent–one

recodedto

havefew

ercategories.B

ecausethese

two

artifactually-related

variablesare

notperfectlycorrelated,they

would

havenot

haveinduced

asingularity

ina

statisticalanalysisand

mighthave

beenundetected

ina

thoughtlessautom

ateddata

mining.

Figure7

shows

ascagnostics

SPLOM

of62

variablesfrom

am

icroarraydataset

[2].The

SPLOM

shows

thehigh

degreeof

homogeneity

inthe

2Dm

arginaldistributionsof

the62

celllinesin

thearray.

Thelinked

scatterplotshow

sa

typical2D

distribu-tion,

evidentlynot

bivariatenorm

al.The

authorsidentified

clus-ters

ofnon-cancerous

andcancerous

cellsin

thesedata.

We

must

remem

berthat

lackof

evidencefor

clustersin

the2D

scattersis

notevidenceforlack

ofclustersin

higherdimensions.Instead,the

scagnosticSPLO

Mlends

some

supporttothe

authors’findingsby

downw

eightingthe

possibilitythatclusters

mightbe

dueto

mea-

surement

artifacts(e.g.,

mixing

discreteand

continuousvariables

inthe

analysis).This

applicationilllustrates

theappropriate

focusofscagnostics

asa

preliminary

screeningm

ethod.Figure

8show

sa

SPLOM

ofw

eatherdata.

Thedata

comprise

hourlym

eteorologicalmeasurem

entsover

ayear

attheG

reenlandH

umboldtautom

aticw

eatherstationoperated

byN

ASA

andN

SF.These

measurem

entsare

partof

theG

reenlandC

limate

Netw

ork(G

C-N

et)sponsored

bythese

federalagencies.

We

havesorted

thevariables

inthe

SPLOM

usingthe

sizeof

theloadings

onthe

firstprincipalcomponentof

thescagnostic

measures.

Thesorting

clearlysegregatesthe

discreteand

continuousvariablesandclusters

similarm

arginal2Ddistributions.

6C

ON

CL

USIO

N

As

we

haveseen,

scagnosticsoffers

thepossibility

ofdetecting

anomalies

inlarge

scatterplotmatrices.

Thereis

more

todo,how

-ever.

First,we

canconstructm

ultivariatem

odelsfrom

theseand

othergraph-theoretic

measures

andapply

themto

them

oregen-

eralpattern

detectionproblem

inhigh-dim

ensionalspace.

Thisapproach

follows

thosetaken

bythe

manifold

learningcom

mu-

nity[32],[4],[6].Second,w

ecan

usethese

methodsto

sortscatter-plotm

atricesin

ordertom

akethem

more

accessibleto

lensingand

L. Wilkinson, A. Anand and R. Grossman, "Graph-theoretic scagnostics," IEEE Symposium on Information Visualization, 2005. INFOVIS 2005., 2005, pp. 157-164.doi: 10.1109/INFVIS.2005.1532142

© Munzner/Möller

multiple keys - partitioning / subdivide

• e.g. 2 keys – use two perpendicular axis OR – use alignment on one axis

• separate by A first and then by B (left) • separate by B first and then by A (right)

• also known as dimensional stacking

93


Overview• Spatial channel

– quantitative vs. categorical attributes – Keys: the importance of ordering

• list (1D) vs. matrix (2D) vs. partition / subdivide (multiple D) – Spatial layout

• rectilinear • parallel • radial

– Spacefilling – Dense

• Linemarks – Connection – Containment

• Using color

© Munzner/Möller

SPloMs

• one scatter plot: choose two dimensions • SPloM: Scatter Plot Matrix

– show all combinations

96

© Munzner/Möller

PC: Axis Ordering

• geometric interpretations – hyper plane, hypersphere – points do have intrinsic order

• nominal / categorical data – no intrinsic order, what to do? – indeterminate/arbitrary order

• weakness of many techniques • downside: human-powered search • upside: powerful interaction technique

• most implementations – user can interactively swap axes

99

© Munzner/Möller

PC vs. scatter plots

• PC – shows only a subset of axis combinations – relatively compact – has a learning curve

• SPloMs – show all combinations of 2 attributes – demanding on screen real-estate – intuitive / known to a broad audience

101

© Munzner/Möller

Radial layouts

• use angular channel for dimensions • rectilinear bar chart vs. radial star plot vs.

radial multipodes plot

102Booshehrian et al, EuroVis 2012

© Munzner/Möller

Connection vs. Containment• relevant for drawing trees • containment -- essentially treemaps • connection -- traditional view of graphs • type of information is different

104Auber: http://tulip.labri.fr/Documentation/3_7/userHandbook/html/ch06.html

http://tulip.labri.fr/Documentation/3_7/userHandbook/html/ch06.html

© Munzner/Möller

Node-link issues

• avoid hairball effect

105http://www2.research.att.com/~yifanhu/GALLERY/GRAPHS/index1.html

http://www2.research.att.com/~yifanhu/GALLERY/GRAPHS/index1.html

© Munzner/Möller

Hyperbolic View

107

3D vs. 2D Hyperbolic Scalability

� information density: 10x better

H3 PARC Tree

fringe

thousands

hundreds

center

dozens

dozens

3D

2D

© Munzner/Möller

Overview• Marks + channels • Channel effectiveness

– Accuracy – Discriminability – Separability – Popout

• Channel characteristics – Spatial position – Colour – Size – Tilt (angle) – Shape (glyph) – Stipple (texture) – Curvature – Motion 109

marks + channelsdap.vsb.cz/cass2017/img/3_thu_0830_vis_encoding... · 2017. 9. 7. · ©...

Documents