infovis approaches -...

InfoVis Approaches

Sheelagh Carpendale

Slides by: Sheelagh Carpendale

Sheelagh Carpendale

Fall 2012

InfoVis Approaches

• Semiotics

• Design

• Perceptual cognitive science

• Data and task

• Usability iteration

• Human activity

• Storytelling


• Storytelling

InfoVis and Semiotics


Semiotics: basic concepts

• Can be thought of as a linguistic problem

• Meaningful representation of data

• Start with fundamental building blocks

• visual variables

• ‘marks’ meaningless in themselves – organized to create (agreed upon) meaning

• Basic source: Jacques Bertin


• Introduction to communication theory (Fiske)

Bertin’s disclaimerhe considers • printable, on white paper, • visible at a glance• reading distance of book or atlas

normal and constant lighting• normal and constant lighting• readily available graphic means


Where does one start?with marks!• for us, pixels?

Visual Variables: how can we vary marks?• by where we place them• by how we place them (Bertin calls this ‘implantation’)• by their visual characteristics (Bertin calls these retinal variables)


The PlanePoints• “A point represents a location on the plane that has no theoretical length or

area. This signification is independent of the size and character of the mark which renders it visible.”

• a location• a location• marks that indicate points can vary in all visual variables

Lines• “A line signifies a phenomenon on the plane which has measurable length

but no area. This signification is independent of the width and characteristics of the mark which renders it visible.”

• a boundary, a route, a connection


Areas • “An area signifies something on the plane that has measurable size. This

signification applies to the entire area covered by the visible mark.”• an area can change in position but not in size, shape or orientation without

making the area itself have a different meaning

Visual Variables


Visual Variablesposition

- changes in the x, y, (z) location

size- change in length, area, repetition

shape - infinite number of shapes

value - changes from light to dark

orientation


orientation - changes in alignment

colour - changes in hue at a given value

texture - variation in pattern

Characteristics of visual variables can be• selective

is a change in this variable enough to allow us to select it from a group?

Visual Variables

g p

• associativeis a change in this variable enough to allow us to perceive them as a

group?

• quantitative is there a numerical reading obtainable from changes in this variable?

• order


are changes in this variable perceived as ordered?

• lengthacross how many changes in this variable are distinctions perceptible?

Semiotics: advantages and challenges

Advantages

• Generative!

Challenges

• Does not in itself include a method for improvement

• In literature, and social science, in general this is done via a critical approach, usually recognized as a sub-di i li ( i i l d )


discipline (e.g. critical pedagogy)

InfoVis and Design


Design: basic concepts

• Draws upon the whole universe as inspiration (biomimicry is part of this)

E cellence thro gh ol me (at least 10 initial sketches• Excellence through volume (at least 10 initial sketches part of this)

• Guidelines

- Essential to know, limiting to follow

• Critique

- Fundamental for success


Fundamental for success

• Main source: Tufte

Tufte’s 1st set of guidelines

What he terms ‘graphical theory’• Show the data

• Induce viewer to think about the substance• Induce viewer to think about the substance

• Avoid distorting the data

• Present many numbers in a small space

• Make large data sets coherent

• Encourage visual comparisons


• Reveal data a several levels of details

• Serve a clear purpose (description, explanation, tabulation, decoration)

• Integrate closely with text descriptions

Graphical Excellence

Start with reasonable data

A. New York stock prices

B. Solar radiation inverted,

C. London stock prices

For all months 1929


Age-adjusted death rates by cancer type for USA

(each some 21,000 numbers)

Can be considered at many levels from overall pattern to county by county detail

• High death rates in north east and around great lakes

• Low rates in band down middle


• Higher rates for men than women in south

• Hot spots; in Minnesota, Iowa, Nebraska, along the Missouri River

• Differences in cancer types by regions

Atlas of Cancer Mortality in the U.S., 1950-94 (Book)

http://www3.cancer.gov/atlasplus/index.html

Data map: 1864 Exports of French Wine


E. Tufte “Visual Display of Quantitative Information” p 25,

Time SeriesE. J. Marey. 1885. Train schedules from Paris to Lyon

Stations spaced according to distances, time from left to right

E.J. Marey, “La Methode Graphique,” (Paris 1885), p.20. This method is attributed to the french engineer, Irby (Tufte, 1883, p.31)


Time SeriesE. J. Marey. 1885. Train schedules from Paris to Lyon

Stations spaced according to distances, time from left to right

E.J. Marey, “La Methode Graphique,” (Paris 1885), p.20. This method is attributed to the french engineer, Irby (Tufte, 1883, p.31)

1981 – new express train – trip now 3 hours instead of 9


Time Series

Diagrams of motion

Using white tape and black velvet, Marey created time series imagescreated time series images.E. J. Marey, (1830 – 1904) E.J. Marey, “Movement,” (London 1895),

p.60 and 61. (Tufte, 1883, p.35-36)


Space-time storySmall multiples

Learn once

Invite comparisons

Los Angeles Times, July 22, 1979; based on work of G. McRae, California Institute of technology. (Tufte, 1983, p.42)


Relational GraphicsRelationship between temperature and thermal conductivity of copper

Gathers dataGathers data from several laboratories

Makes a clearer and stronger point by the collection

Connected points are from one publication

Different answers


Different answers result from different impurities levels

Graphical Excellence - Summary

Designed for the presentation of interesting data – matter of substance, of statistics, and of design.

Graphical excellence consists of complex ideas communicated with clarity, precision and efficiencyprecision and efficiency.

Graphical excellence is that which gives to the viewer the greatest number of ideas in the shortest time with the least ink in the smallest space.

• > ideas• < time• < ink• < space


Graphical excellence is nearly always multivariate.

And is all about truth and integrity.

Graphical integrity

• Graphics can be a powerful communication tool

• Lies and falsehoods are possible

• Much focus on this ‘how to lie with maps’ or ‘statistics’


Examples of misleading graphics


Where is the bottom line? What is happening in 1970?Day Mines, Inc. 1974 Annual Report, p1 (Tufte, 1983, p54)

Misleading graphics New York Times, August 8, 1978, p.D-1 (Tufte, 1983, p54)

What is the first impression of the airlines relative success in 1978?

Pittsburgh Civic Commission, Report on Expenditures of the Department of Charities (Pittsburgh, 1911), p.7 (Tufte, 1983, p54)

Order of numbers?

Magnitude of numbers? Impression?


Achieving graphical Integrity

A graphic does not distort if the visual representation is consistent with the numerical representation.

•Is the magnitude of ‘visual representations’ as physically measured on the graphic?graphic?

•Or the perceived magnitude?

Approach

Conduct a study of visual perception of the graphics.

Circles – perceived area grows more slowly than measured area


reported perceived area = (actual area)X, where x = 0.8+/-0.3

Lines -

Lie Factors

•‘the representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the numerical quantities represented.’

•‘Clear, detailed and thorough labeling should be used to defeat graphical distortion and ambiguity. Write out explanations of the data on the graphic itself. Label important events in the data.’

Lie Factor = size of effect shown in graphic


– Lie factor of 1 – is desirable

– lie factors > 1.05 or < 0.95 go beyond plotting errors

size of effect in data

Fuel economy standards for automobiles18 miles/gallon in 1978 to 27.5 miles/gallon in 1985Increase of 53% = (27.5 – 18.0)/(18.0) x 100

Extreme example


Extreme example

Graphic increase

783% = (5.3 – 0.6)/(0.6) x 100

Lie Factor = 783/53 = 14.8

Additional confounding factors


gUsually the future is in front of usDates remain same size and fuel factors increase Includes perspective distortion – how to read change in

perspective

Extrapolationa graphic generates visual expectations – deception can

result from incorrect extrapolation of visual expectations

1st seven intervals areintervals are 10 years

The last interval is 4 years

Gives a false sense of decline


National Science Foundation, Science Indicators, 1974 (Washington D.C., 1976), p.15, (Tufte, 1983, p60)

Accurate data for the next 10 years

Design Variation vs Data Variation

New York Times, Dec. 19, 1978, p.D-7 (Tufte, 1983, p61)


Context is Essential

Different data points would tell a different stories

Graphics must not quote data out of context

a different stories


Context is Essential

Graphics must not quote data out of context

Comparisons with adjacent states give more context


Graphical Integrity - Summary

• ‘The representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the numerical quantities represented.

• Clear, detailed, and thorough labeling should be used to defeat graphical distortion and ambiguity. Write out explanations of the graphic itself. Label important events.

• Show data variation, not design variation.

• In time-series displays of money, deflated and standardized units of monetary measurement are nearly always better than nominal units.

• The number of information carrying (variable) dimensions depicted


• The number of information-carrying (variable) dimensions depicted should not exceed the number of dimensions in the data.

• Graphics must not quote data out of context.’

(Tufte, 1983, p77)

Tufte’s Graphical Theory

Tutfe presents • 1st graphical excellence• 2nd graphical disasters

Then discusses the causes and provides guidelines1. Causes of poor graphics2. Guidelines

1. Minimize ink while maximizing data ink2. Data density3. Avoid chartjunk


j4. Create multifunctioning graphics 5. Make use of parallelism (small multiples)6. Consider 3D carefully 7. Consider colour carefully

Causes of Poor Graphics

Question: why are there so few examples of good graphics?

• One possible answer is lack of training

• Another is societal attitudes

Common Attitudes

• Data is boring – in spite of the fact that more and more time, money, and people hours are spent gathering data (fire hoses of data)

• Graphics are for people who can not understand the data

Consequences

• Simplification


Simplification

• Decoration

• Unnecessary emphasis

• And LIES

Guidelines: data density

Data density = number of entries in the data matrix / area of the data graphic

Data can be designed to have several viewing depths

1 What is seen from a distance an overall structure usually aggregated from an1. What is seen from a distance, an overall structure usually aggregated from an underlying microstructure• The overall pattern the concentrations

2. What is seen close up and in detail; the fine structure of the data• The size etc. of the cities • The spread in the sparse areas

3. What is seen implicitly – the interrelations between the data • The effect of traffic and landscape corridors


The effect of traffic and landscape corridors

Guidelines: data densityNew York Weather History- 1980• Data density - 181 numbers/sq inch


New York Times, (Jan. 1981), p.32, (Tufte, 1883, p.30)

Guidelines: Avoid chartjunk

Adding frills does not help

Data graphics get sold on their content

Avoid decoration

Moiré vibration

Over powering grids

Extraneous additions


Guidelines: Avoid chartjunk

Cotton production in Brazil, 1927 – vibrating textures


Chart Junk: A common error

Information visualization is not just pretty graphics• graphical re-design by amateurs on computers gives us

- “fontitis,” “chart-junk,” etc.

10

5

8

6

8

10Dear Sir; This is a really exciting opportunity! Take advantage of it !


2

0

2

4

Guidelines: Create multifunctioning graphics

Reading chart vertically ranks 15Reading chart vertically – ranks 15 countries by government tax collection for 1970 and 1979

Names are spaced in proportion to percentages


Paired comparisons show how numbers changed over the years

Guidelines: Make use of parallelism

Often there exists some parallelism in the data

This can be used do clarify which aspects of the data changechange

Small multiples have been used for 100s of years• These are a series of graphics such as frames in a movie

showing the same combination of variables, indexed by changes in another variable(s)

Well designed small multiples are


g• Inevitably comparative• Deftly multivariate• Shrunken, high density graphics• Usually based on a large data matrix• Drawn almost entirely with data-ink• Efficient in interpretation (learn once applies to all)• Often narrative in content – showing shifts in relationships


Small Multiple from Huygens’ Systema Saturnium 1659.


Inner ellipse is earth’s orbit, outer Saturn’s orbit, outer most floating images depict Saturn as viewed from earth, explaining previous idea about Saturn’s shape



J. H. Colton

Johnson’s New Illustrated Family Atlas with Physical Geography, 1864

Small Multiples: Showing Time and Change


Small Multiples: Showing Time and Change


Guidelines: Consider 3D carefully

Why 3D?

• The world is 3D – 2D graphics flatten data


Guidelines: Consider 3D carefully

3D graphic excellence


Guidelines: Consider 3D carefully 3D graphic excellence

Some 7000 pieces of space debris –points not to scale pbut each at least as big as 8x10 inches – 1987

Micro and macro readings


Guidelines: Consider 3D carefully 3D graphic excellence

3D explosion diagram IBM Series III copier


Guidelines: Consider colour carefully

To avoid over powering and garish colours


Guidelines: Consider colour carefully Use least possible emphasis colours

on clam background


Guidelines: Consider colour carefully

To avoid over powering and garish colours

Use least possible distinguishable difference


Guidelines: narratives of space and time Use least possible

distinguishable difference


Guidelines: narratives of space and time Use least possible distinguishable

difference


Graphical Integrity - Summary

• ‘The representation of numbers, as physically measured on the surface of the graphic itself, should be directly proportional to the numerical quantities represented.

Cl d il d d h h l b li h ld b d d f• Clear, detailed, and thorough labeling should be used to defeat graphical distortion and ambiguity. Write out explanations of the graphic itself. Label important events.

• Show data variation, not design variation.

• In time-series displays of money, deflated and standardized units of monetary measurement are nearly always better than nominal units.


• The number of information-carrying (variable) dimensions depicted should not exceed the number of dimensions in the data.

• Graphics must not quote data out of context.’

(Tufte, 1983, p77)

Design: advantages and challenges

Advantages

• Lots of guidelines

• Lots of examples

• A practice of critique

Challenges

• No place to start


• No place to start

• Lots of guidelines

• Lots of examples

InfoVis and Perceptual Cognitive Science


Perceptual cognitive science: basic concepts

• Understanding perceptual cognitive science

• Well established empirical methods with which to assess percept al iss esassess perceptual issues

• Active research community to monitor


ReferencesColin Ware. (2004) Information Visualization: Perception for Design. Morgan

Kaufmann.

Maureen Stone. (2003) A field guide to digital color. AK Peters

S S St (1961) Th P h h i f S F ti SS. S. Stevens. (1961) The Psychophysics of Sensory Function. Sensory Communication, MIT Press, pp 1-33.

William S. Cleveland, Robert McGill. (1984) Graphical Perception: Theory, Experimentation and the Application to the Development of Graphical Models. J. Am. Stat. Assoc. 79:387, pp. 531-554.

Leland Wilkinson. (1993) Comment on Cleveland. Journal of Computational and Graphical Statistics, 2, pp. 355-360.


Bernice E. Rogowitz and Lloyd A. Treinish. (1996) How Not to Lie with Visualization. Computers In Physics 10(3), pp 268-273. http://www.research.ibm.com/dx/proceedings/pravda/truevis.htm

C. Brewer. (1999) Color use guidelines for data representation. http://www.personal.psu.edu/faculty/c/a/cab38/ColorSch/ASApaper.html

Field of ViewUseful field of view

varies with tasklow character density - as wide as 15o

high character density - as narrow as 1o to 4o


Saccades

Fovea gives small high resolution images

Saccades do rapid scanning

Brain assembles

Vision perceived as continuous


Eye movementsSaccadic movements• eye moves rapidly from fixation to fixation• dwell period 200 to 600 msec• saccade takes 20 to 100 msecsaccade takes 20 to 100 msec• peak velocity can be 900 deg/sec• ballistic - cannot be adjusted mid saccade• saccadic suppression - less sensitive visually during a saccade

smooth-pursuit movements• ability to ‘lock-on’ to a smoothly moving object• enables head and/or body movements while maintaining visual

contact


contact

convergent movements• towards - eyes converge • away - eyes diverge

accommodation• new target - refocus - 200 msec

d d ti l i ll l d

Image from John MCannImage from John MCann (slide M. Stone)Image from John MCann (slide M. Stone)


Image from John MCannImage from John MCann


Image from John MCann (slide M. Stone)Image from John MCann (slide M. Stone)


ConsequencesCrispening


ConsequencesCraik-Cornsweet Effect


Reading and short term memoryHow many symbols can you remember?

• Usually about 7• 7+ or - 2• short term

memory

as 1o

X?

#

Q

@6$


&%

Q

9*F

Pre-attentive processing23589457397568607967524535123465346243562457624572456134523523523523523524351345324716498762987460329587235827653363787213876429876987636409872169653296241392374621639876398712365971245938746387469887126498172649872165971523972356987129721653978216409871298172649872165971523972356987129721653978216409871246478346721898763945089776439821734694649643927643098726342874698649875971523971239764908714698764987243698127346987461435895321456865437

2358945739756860796752453512346534624356245762457245613452352352352352352435134532471649876298746032958723 82 6 3363 8 2 38 6 298 698 636 098 2 696 32962 3


235827653363787213876429876987636409872169653296241392374621639876398712365971245938746387469887126498172649872165971523972356987129721653978216409871246478346721898763945089776439821734694649643927643098726342874698649875971523971239764908714698764987243698127346987461435895321456865437

Pre-attentive processing

1000

typical results

X

Response time(milliseconds)

750

500 X X X


Number of distracters

250

9 1263pre-attentivenon-pre-attentive

X

Pre-attentive processing

orientation shapecurved/straight

numbersizeshape


Position: best for all data typesdata

type


Mackinlay, Automating the Design of Graphical Presentations of Relational Information, ACM TOG 5:2, 1986

Based on visual gestalt (perception of ‘wholeness’)

Gestalt Principles: perception

Proximity Similarity


Continuity

Gestalt Principles


Information Visualization: Perception for Design. Ware, Morgan Kaufmann, 2000

connectedness

Gestalt Principles



Closure

Gestalt Principles


Closure

O l i it

Gestalt Principles

Overrules proximity, similarity



Symmetry

Emphasizes relationships

Gestalt Principles

p p



Figure/ground

Gestalt Principles


Smaller components seen as objects

Colour The Retina


photoreceptors: rods and cones

neurons (receptive fields): intermediate neural layers – image processing

http://school.discovery.com/homeworkhelp/worldbook/atozpictures/lr001100.htmlhttp://www.ccrs.nrcan.gc.ca/ccrs/eduref/sradar/chap2/c2p2_g2e.html

Colour Trichromacy theory

We have 3 types of cones, or colour receptors

Therefore, most colour systems have 3 dimensions :

paint red yellow bluepaint red, yellow, bluetv red, green, blueprinter cyan, magenta, yellow


Colour Blindnesssmall-field tritanopia


Colour Opponent Process Theory

Late 19th C Ewald Hering

6 elementary coloursy

arranged perceptually as opponent pairs along 3 axes:black – whitered – greenblue – yellow


cornerstone of modern colour theory

well established phsyiological basis

Colour Opponent Process Theory - continued

Many lines of scientific evidence worth examining

☼ Namingg

☼ Cross-Cultural naming

☼ Unique Hues

☼ Neurophysiology

☼ Categorical colours


colour constancy

colour perception


Colouring categorical data

• limited number readily distinct (spatially separate colour patches)

• think about selection, association, and adjacent distinction• Ware’s maximally discriminable colours


Colour scales (maps)

Rainbow (hue)No ordering- No ordering

- Good name space

Greyscale/luminance/saturation

- ordered


http://www.research.ibm.com/dx/proceedings/pravda/index.htmhttp://www.research.ibm.com/dx/proceedings/pravda/index.htmhttp://www.research.ibm.com/visualanalysis/perception.htmlhttp://www.research.ibm.com/visualanalysis/perception.html

Rainbow scale

- No ordering - Good name space (green part, yellow part ….)

- Jet engine noise simulation



Two-Hue scale

- Easier to see small variations



Perceptual cognitive science: advantages and challenges

Advantages

• Provides a science on which to base choices• Provides a science on which to base choices

• Provides perceptual guidelines

• Basing choices on human perception

Challenges

• Requires research and extrapolation


• Not prescriptive

• Rapidly changing field of research

• Does not consider learning

InfoVis and Data and Task Approach


Data and task: basic concepts

• Understand the data

• Understand the tasks

• Work with a subset (perhaps one) of the tasks

• Relates to HCI and requirements engineering approaches

• Best resources

• task-based interface design


• Requirements engineering literature

• *** Amar and Stasko’s paper

Data and task: advantages and challenges

Advantages

• Practical approach

• Advice from HCI and SE

Challenges

• Does not include any representational basis

• Stays with the understood


• Stays with the understood

InfoVis and Usability Iteration


Usability iteration: basic concepts

• Process

- Make something

- Run a usability study

- Improve initial attempt

- iterate



Early design

Early usability evaluation can kill a promising idea• focus on negative ‘usability problems’

ideaidea idea

idea


idea

Early designsIterative testing can promote a mediocre idea

idea1idea1

idea1

idea1

idea1


Early designGenerate and vary ideas, then reduce

idea5

idea4

idea3

idea2 idea5

idea6idea7

idea8

idea9

idea5

idea1


evaluate the better ideas

idea41

Early designs as sketchesGetting the design right

Getting the right design

idea1idea1

idea1

idea1

idea1


idea5

idea4

idea3

idea2 idea5

idea6idea7

idea8

idea9

idea5

idea1

Usability iteration: advantages and challenges

Advantages

• Refines a given piece of software

• Widely accepted

• Belongs in industry

Challenges

• Can kill a brilliant idea


• Can kill a brilliant idea

• Can promote mediocrity

• Does not generalize

InfoVis and Human Activity


Human activity: basic concepts

• Observation for design

• Ethnographically inspired methodologies available

• Activity theory

• Little practiced currently


Observation for design

Motivation – why observe

Observation as a skill – how to learn/teach

Observation for new insight – how to catch that ‘edge’


Henry Ford

‘If I had asked people what they wanted, they would have said faster horses’


observation for inspiration


informed by ethnographic methods


ethnographic studies done professionally


Ethnographic studies done professionally

not the right settings

no intention for invention


we have an

ethnographically inspired approach


with intention for invention


or possibility intervention


heavily based on observation


observation is a skill


Other observation possible levels

Lower level

Meta level

Finer details

Kinetic level

….


Maybe multiple possible ‘levels’

problem is we see what we expect to see


relates to change blindness


http://www.youtube.com/watch?v=Ahg6qcgoay4&NR=1

http://www.youtube.com/watch?v=diGV83xZwhQ&feature=fvsr


Change Blindness

some times changes are not perceived

http://viscog.beckman.uiuc.edu/djs_lab/demos.html

http://www.psych.ubc.ca/~rensink/

Rensink, Ronald A.; O'Regan, J. Kevin & Clark, James J. (1997), To see or not to see: the need for attention to perceive changes in


see or not to see: the need for attention to perceive changes in scenes, Psychological Science 8 (5): 368-373.

Silverman, M. & Mack, A. (2006), Priming by change blindness: When it does and does not occur, Consciousness and Cognition 15: 409-422.

Simons, Daniel J. & Levin, Daniel T. (1998), Failure to detect changes to people during a real-world interaction, Psychonomic

the challenge

to see beyond our expectations to reality


bottom line

unless we work at it we see what we expect to see


Albert Einstein

‘Everything that can be counted does not necessarily count; everything that counts cannot necessarily be counted’


Human activity: advantages and challenges

Advantages

• Rich ‘slice’

• New understanding

• Theoretical insight

• Inspiration

Challenges


Challenges

• Hard to ‘catch the edge’ – to see beyond the expected

• A lot of work

• Offers insight not design

InfoVis and Storytelling


Storytelling: basic concepts

• New direction

• Visualization as a communication medium

• On the web

• In the news

• New perspective based in infographics


Storytelling: advantages and challenges

Advantages

• Reaching wider audience

• New perspectives bring new understandings

• Learning from infographics

Challenges

• Shows possibility for bias


• Shows possibility for bias

• Requires a critical response

InfoVis Approaches

• Semiotics

• Design

• Perceptual cognitive science

• Data and task

• Usability iteration

• Human activity

• Storytelling


• Storytelling

infovis approaches -...

Documents