CS 791/891 Seminar Fall 2015

Intro to Info Vis

Dr. Michele C. Weigle

http://www.cs.odu.edu/~mweigle/CS891-F15/

Outline

! Motivation -- why vis?

! Tools

! What, Why, How Framework

! How ! marks and channels ! types

2 CS 791/891 - Fall 2015 - Weigle What is visualization? ! "The communication of using graphical representations" ! Ward, Grinstein, Keim

! "The use of computer-supported interactive visual representations of data to amplify cognition" ! Card, Mackinlay, Shneiderman, Readings in Information Visualization: Using Vision to Think

! "The purpose of visualization is insight, not pictures." !

3 CS 791/891 - Fall 2015 - Weigle

Where have you seen a visualization today?

http://hamptonroads.com/weather http://hamptonroads.com/traffic

4 CS 791/891 - Fall 2015 - Weigle What's vis?

! Visualization is suitable when there is a need to augment human capabilities rather than replace people with computational decision-making methods.

! The design space of possible vis idioms is huge, and includes the considerations of both how to create and how to interact with visual representations.

! Vis design is full of tradeoffs, and most possibilities in the design space are ineffective for a particular task.

! Vis usage can be analyzed in terms of why the user needs it, what data is shown, and how the idiom is designed.

Munzner, pg. 1 5 CS 791/891 - Fall 2015 - Weigle

augment human capabilities Why have a human in the loop?

! Vis allows people to analyze data when they don't know exactly what questions to ask in advance.

! Best path - put a human in the loop

! exploit the pattern detection properties of human vision

6 CS 791/891 - Fall 2015 - Weigle augment human capabilities Humans are great at pattern recognition

Create visualizations that lets computers do what computers do well and lets humans do what humans do well.

7 CS 791/891 - Fall 2015 - Weigle

augment human capabilities Why have a computer in the loop?

Munzner, Fig 1.2 (Barsky et al., 2007)

8 CS 791/891 - Fall 2015 - Weigle augment human capabilities Why use an external representation?

! Vis allows people to offload cognition and memory usage to make space for other operations.

! as external representations ! information can be organized by spatial location ! search - grouping items needed for problem-solving in one location ! recognition - grouping relevant info for one item in the same location

9 CS 791/891 - Fall 2015 - Weigle

augment human capabilities Visualization can extend your memory

What is 57 x 48? paper mental buffer 52 57 [8 * 7 = 56] x 48 [8 * 5 = 40 + 5 = 45] 1 45 6 [4 * 7 = 28] 22 8 [4 * 5 = 20 + 2 = 22] 2 7 3 6 [8 + 5 = 13] [4 + 2 + 1 = 7]

Example courtesy Tamara Munzner, Univ. of British Columbia

10 CS 791/891 - Fall 2015 - Weigle augment human capabilities Why depend on vision? ! Visual system provides a high-bandwidth channel to our brains.

! Significant amount of visual information processing occurs in parallel at the pre- conscious level.

11 CS 791/891 - Fall 2015 - Weigle

augment human capabilities Can you find the red dot?

preattentive processing

http://www.csc.ncsu.edu/faculty/healey/PP/index.html

12 CS 791/891 - Fall 2015 - Weigle augment human capabilities

Which state had the highest marriage rate?

Florida - 7.5 Connecticut - 5.9 Colorado - 6.8 Delaware - 5.4 District of Columbia - 4.7

13 CS 791/891 - Fall 2015 - Weigle

augment human capabilities

Which state had the highest marriage rate?

U.S. Census Bureau, Statistical Abstract of the United States: 2012 http://www.census.gov/compendia/statab/2012/tables/12s0133.pdf, http://www.census.gov/compendia/statab/2012/tables/12s0133.xls 14 CS 791/891 - Fall 2015 - Weigle augment human capabilities Why show the data in detail?

! Vis tools can allow people to explore data to find patterns or to determine if a statistical model actually fits the data

! Look out for questionable data ! "just because it's numbers doesn't mean it's true" ! is it a typo or something interesting? ! "make sure you know which one it is"

i i 15 CS 791/891 - Fall 2015 - Weigle i i

8 1. What’saugment Vis, human and Why capabilities Do It? Anscombe's Quartet Anscombe’s Quartet: Raw Data I II III IV xyxyxyxy 10.08.04 10.09.14 10.07.46 8.06.58 8.06.95 8.08.14 8.06.77 8.05.76 13.07.58 13.08.74 13.012.74 8.07.71 9.08.81 9.08.77 9.07.11 8.08.84 11.08.33 11.09.26 11.07.81 8.08.47 14.09.96 14.08.10 14.08.84 8.07.04 6.07.24 6.06.13 6.06.08 8.05.25 4.04.26 4.03.10 4.05.39 19.012.50 12.010.84 12.09.13 12.08.15 8.05.56 7.04.82 7.07.26 7.06.42 8.07.91 5.05.68 5.04.74 5.05.73 8.06.89 mean 9.07.5 9.07.5 9.07.5 9.07.5 var. 10.03.75 10.03.75 10.03.75 10.03.75 corr. 0.816 0.816 0.816 0.816 F.J. Anscombe, "Graphs in Statistical Analysis", American Statistician, Feb 1973 Munzner, Figure 1.3 16 CS 791/891 - Fall 2015 - Weigle

Figure 1.3: Anscombe’s Quartet is four datasets with identical simple statistical properties: mean, variance, correlation, and linear regression line. However, visual inspection immediately shows how their struc- tures are quite di↵erent. After [Anscombe 73], Figures 1–4. From http://en.wikipedia.org/wiki/File:Anscombe.svg

i i i i augment human capabilities The four data sets are not the same

"Graphics reveal data" - , The Visual Display of Quantitative Information http://en.wikipedia.org/wiki/File:Anscombe.svg

17 CS 791/891 - Fall 2015 - Weigle

Why focus on tasks?

! Vis usage can be analyzed in terms of why the user needs it, what data is shown, and how the idiom is designed.

! The intended task is just as important as the data to be visualized.

! Four categories of tasks ! presentation ! discovery ! enjoyment of information ! producing more information for later use

18 CS 791/891 - Fall 2015 - Weigle Outline

! Motivation -- why vis?

! Tools

! What, Why, How Framework

! How ! marks and channels ! chart types

19 CS 791/891 - Fall 2015 - Weigle

Tools of the Trade

http://selection.datavisualization.ch/ 20 CS 791/891 - Fall 2015 - Weigle Excel

http://chandoo.org/wp/2008/09/03/6-charts-to-never-use/

http://www.juiceanalytics.com/writing/recreating-ny-times-cancer-graph/

21 CS 791/891 - Fall 2015 - Weigle

http://www.r-project.org

22 CS 791/891 - Fall 2015 - Weigle Tableau http://www.tableausoftware.com

23 CS 791/891 - Fall 2015 - Weigle

D3

http://d3js.org

24 CS 791/891 - Fall 2015 - Weigle Outline

! Motivation -- why vis?

! Tools

! What, Why, How Framework

! How ! marks and channels ! chart types

25 CS 791/891 - Fall 2015 - Weigle

Analysis: What, why, and how Analysis: What, •why, what and is shown? how • what is shown? – data abstraction • why is the user looking at it? – data abstraction – task abstraction • why is the user •looking how is atit shown?it? – task abstraction – idiom: visual encoding and interaction • how is it shown?• abstract vocabulary avoids domain-specific terms – idiom: visual encoding and interaction Main reference: Visualization Analysis and Design, Munzner. AK Peters / CRC Press, Oct 2014. • abstract vocabulary avoids domain-specific termshttp://www.cs.ubc.ca/~tmm/courses/533/book/

– translation processSlides initerative, this style are tricky shamelessly borrowed from Tamara Munzner's VAD minicourse, June 2014 (http://www.cs.ubc.ca/~tmm/talks.html#minicourse14) 26 • what-why-how analysis framework as scaffold to think systematically about design space

15 What? Datasets Attributes

What? Data Types Attribute Types Items Attributes Links Positions Grids Categorical

Data and Dataset Types Why? Tables Networks & Fields Geometry Clusters, Ordered Trees sets, lists Ordinal Items Items (nodes) Grids Items Items Attributes Links Positions Positions How? Attributes Attributes Quantitative

Dataset Types Ordering Direction Tables Networks Fields (Continuous) Sequential Attributes (columns) Grid of positions

Link Items Cell (rows) Node Diverging (item) Cell containing value Attributes (columns)

Value in cell Cyclic Multidimensional Trees

Value in cell

Geometry (Spatial) Dataset Availability Static Dynamic

Position 16 2716

DatasetDataset types types

Dataset Types

Tables Networks Fields (Continuous) Geometry (Spatial)

Attributes (columns) Grid of positions

Link Items Cell (rows) Position Node (item) Cell containing value Attributes (columns)

Value in cell Multidimensional Table Trees

Value in cell

2817 DatasetDataset andand data data types types

Data and Dataset Types Tables Networks & Fields Geometry Clusters, Trees sets, lists Items Items (nodes) Grids Items Items Attributes Links Positions Positions Attributes Attributes

Data Types Items Attributes Links Positions Grids Dataset Availability Static Dynamic

29 18

AttributeAttribute types types

Attribute Types Categorical Ordered Ordinal Quantitative

Ordering Direction Sequential Diverging Cyclic

30 19 • {action, target} • pairs{action, target} pairs –– discover distribution distribution –– compare trends trends –– llocate outliers outliers –– browse topology topology

31 20

High-level actions: Analyze High-level actions: Analyze consume • Analyze • consume– discover vs –discoverpresent vs present Consume Discover Present Enjoy •• classicclassic split split •• akaaka explore explore vs explainvs explain – enjoy •• newcomernewcomer Produce •• akaaka casual, casual, social social Annotate Record Derive

tag •• produce – annotate, record record – derive –derive • crucial design • crucialchoice design choice

32

21 Actions: Mid-level search, low-level query • what does user know? – target, location

• how much of the data matters? – one, some, all

33

Why: Targets

34 Outline

! Motivation -- why vis?

! Tools

! What, Why, How Framework

! How ! marks and channels ! chart types

35 CS 791/891 - Fall 2015 - Weigle

36 Definitions:Definitions: MarksMarks and and channels channels

• marks• marks Points Lines Areas – geometric– geometric primitives primitives • channels• channels control appearance of marks – – control appearance of marks Position Color – can redundantly code with – can redundantly code with multiple Horizontal Vertical Both multiple channels channels • interactions • interactions – point marks only convey position;– point marks no area only conveyconstraints position; no Shape Tilt • canarea be constraints size and shape coded• can be size and shape coded – line– linemarks marks convey convey positionposition and and length length• can only be size coded in 1D • can(width) only be size coded in Size 1D (width) – area marks fully constrained Length Area Volume – area marks fully constrained • cannot be size or shape coded • cannot be size or shape 31 coded 37

Visual encoding Visual encoding • analyze idiom structure • analyze– as combination idiom structure of marks and channels – as combination of marks and channels

1: 2: 3: 4: 1:vertical position 2:vertical position 3:vertical position 4:vertical position vertical position verticalhorizontal position position verticalhorizontal position position verticalhorizontal position position horizontal position horizontalcolor hue position horizontalcolor hue position color hue colorsize (area)hue size (area) mark: line mark: point mark: point mark: point

mark: line mark: point mark: point mark: point 38 32 Channels: Expressiveness types and effectiveness rankings

39

Effectiveness and expressiveness principles • effectiveness principle – encode most important attributes with highest ranked channels

• expressiveness principle – match channel and data characteristics [Automating the Design of Graphical Presentations of Relational Information. Mackinlay. ACM Trans. on Graphics (TOG) 5:2 (1986), 110–141.]

• rankings: where do they come from? – accuracy – discriminability – separability – popout

40 Accuracy: Vis experiments

[Crowdsourcing Graphical Perception: Using Mechanical Turk to Assess Visualization Design. Heer and Bostock. Proc ACM Conf. Human Factors in Computing Systems (CHI) 2010, p. 203–212.] after Michael McGuffin course slides, 41 http://profs.etsmtl.ca/mmcguffin/

Discriminability: How many usable steps? • linewidth: only a few

[mappa.mundi.net/maps/maps 014/telegeography.html] 42 SeparabilitySeparability vs. Integralityvs. Integrality

Position Size Width Red Hue (Color) Hue (Color) Height Green

Fully separable Some interference Some/signifcant Major interference interference

22 groupsgroups each each 22 groupsgroups each each 33 groupsgroups total: total: 44 groupsgroups total: total: integralintegral area area integralintegral hue hue

4338

PopoutPopout / Preattentive Processing •• findfindfind thethe the redred red dotdot dot – how long does it take? ––howhow longlong doesdoes itit take?take? •• parallelparallel processingprocessing processing onon manymany on manyindividualindividual channelschannelsindividual channels ––speed speedspeed independentindependent independent ofof distractordistractor of distractor countcount count ––speed speedspeed dependsdepends depends onon channelchannel on channel andand amountamount and ofof amount of difference from distractors differencedifference fromfrom distractorsdistractors •• serialserial searchsearch search forfor (almostfor(almost (almost all)all) combinationscombinations all) combinations––speedspeed dependsdepends onon numbernumber ofof distractorsdistractors – speed depends on number of distractors

443939 3939 GroupingGrouping

• containment• containment • connection • connection

• proximity Identity Channels: Categorical Attributes –• sameproximity spatial region Spatial region • similarity– same spatial region – same values as other Color hue •categoricalsimilarity channels – same values as other Motion categorical channels

Shape 45 41

Relative vs. absolute judgements RelativeRelativeRelative vs. vs. absolute vs. absolute absolute judgements judgements judgments • perceptual•• perceptual system system system mostly mostly mostly operates operates operates withwith relativerelative with judgements, relative judgments,notnot absoluteabsolute • perceptualnot absolute system mostly operates with relative judgements, not absolute – that’s– that’s why why accuracy accuracy increases increases with with common common frame/scaleframe/scale and alignment – that’s– that why’s whyaccuracy accuracy increases increases with common with frame/scale common andframe/scale alignment and alignment – Weber’s– Weber’s Law: Law: ratio ratio of of increment increment to to background background isis constantconstant – Weber’s– Weber Law:’s Law:ratio ofratio increment of increment to background to background is constant is constant • filled• filled rectangles rectangles differ differ in in length length by by 1:9, 1:9, difficult difficult judgement judgement • filled• filled rectangles rectangles differ in differ length inby length1:9, difficult by 1:9,judgement difficult judgment • white• white rectangles rectangles differ differ in in length length by by 1:2, 1:2, easy easy judgement judgement • white• white rectangles rectangles differ in differ length inby length 1:2, easy by judgement 1:2, easy judgment

B B A A A B lengthlength positionpositionFramed along along positionposition along length positionunalignedunaligned along positionalignedaligned alongscale unalignedcommon scale aligned scale common scale 42 after [Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods. Cleveland and McGill. Journ. American Statistical Association 79:387 (1984), 531–554.] 42 46 afterafter [Graphical [Graphical Perception: Perception: Theory, Theory, Experimentation, Experimentation, and Application and Application tocommon the Development to the Development of scale Graphical Methods. of Graphical Cleveland Methods. and McGill. Cleveland Journ. American and McGill. Statistical Journ Association. American 79:387 Statistical (1984), 531–554.] Association 79:387 (1984), 531–554.] 42 after [Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods. Cleveland and McGill. Journ. American Statistical Association 79:387 (1984), 531–554.] Outline

! Motivation -- why vis?

! Tools

! What, Why, How Framework

! How ! marks and channels ! chart types

47 CS 791/891 - Fall 2015 - Weigle

48 ArrangeArrange spacespace Encode

Arrange Express Separate

Order Align

Use

49 4

Arrange tables tables

Express Values Axis Orientation Rectilinear Parallel Radial

Separate, Order, Align Regions Separate Order Layout Density Dense Space-Filling Align

1 Key 2 Keys 3 Keys Many Keys List Matrix Volume Recursive Subdivision

505 KeysKeys and and valuesvalues Keys and values Tables • key • key Attributes (columns) – independent• key attribute – independent attribute Items – used as– uniqueindependent index attribute to look up items (rows) – used as unique index to look up items – simple tables:– used as 1 unique key index to look up items – simple tables: 1 key Cell containing value – multidimensional– simple tables: tables: 1 key multiple keys – multidimensional tables: multiple keys • value – multidimensional tables: multiple keys Multidimensional Table –• dependentvalue• value attribute, value of cell • classify– dependent arrangements attribute, value ofby cell key count – dependent attribute, value of cell – 0, 1, 2, many... Value in cell • classify• classify arrangements arrangements by key bycount key count – 0, 1, 2, many... – 0, 1, 2, many... Express Values 1 Key 2 Keys 3 Keys Many Keys List Matrix Volume Recursive Subdivision

51 6 6

Idiom:Idiom: scatterplotscatterplot Express Values • express• express values – quantitative– quantitative attributes • no keys, only values • no keys, only values – data –• data2 quant attribs – mark:• 2 quant points attribs – channels– mark: points –• channelshoriz + vert position tasks – • horiz + vert position • find trends, outliers, – tasksdistribution, correlation, clusters• find trends, outliers, distribution, correlation, clusters – scalability– scalability • hundreds of items • hundreds of items [A layered grammar of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 3–28.] [A layered grammar of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 3–28.] 52 7 Time Series Data Animated Scatterplots ! When to use: ! to compare how two quantitative variables change over time

http://www.gapminder.org/videos/200-years-that-changed-the-world-bbc/

Full TED talk (20 min): http://www.ted.com/index.php/talks/hans_rosling_shows_the_best_stats_you_ve_ever_seen.html 53 CS 791/891 - Fall 2015 - Weigle Few, Now You See It

Some keys: Categorical regions SomeSome keys: keys: Categorical Categorical regions regions

Separate Order Align

• regions: contiguous bounded areas distinct from each other • regions: contiguous bounded areas distinct from each other –•usingregions space : tocontiguous separate (proximity)bounded areas distinct from each other ––using using space space to separate to separate (proximity) (proximity) – following– following expressiveness expressiveness principle forprinciple categorical for attributescategorical attributes – following expressiveness principle for categorical attributes • use• use ordered ordered attribute attribute to order to andorder align and regions align regions • use ordered attribute to order and align regions 1 Key 2 Keys 3 Keys Many Keys List Matrix Volume Recursive Subdivision

8

548 Idiom: bar chart Idiom: bar chart 100 100 • one key, one value 75 75 • one key, one value 50 50 – data – data 25 25 • 1 categ attrib, 1 quant attrib 0 0 • 1 categ– mark: attrib lines, 1 quant attrib – channels – mark: lines Animal Type Animal Type • length to express quant value channels – • spatial regions: one per mark • length to –expressseparated horizontally, quant valuealigned vertically – ordered by quant attrib • spatial regions:» by label one (alphabetical), per mark by length attrib (data-driven) – separated– task horizontally, aligned vertically – ordered• compare, by quant lookup attrib values »– byscalability label (alphabetical), by length attrib (data-driven) • dozens to hundreds of levels for key attrib – task 9 • compare, lookup values – scalability • dozens to hundreds of levels for key attrib

55

Fig 7.14, p. 152, Few, Now You See It Bar Graphs

! When to use: ! When you want to support the comparison of individual values

56 CS 791/891 - Fall 2015 - Weigle Idiom: stacked bar chart • one more key – data • 2 categ attrib, 1 quant attrib – mark: vertical stack of line marks • glyph: composite object, internal structure from multiple marks – channels [Using Visualization to Understand the Behavior • length and color hue of Computer Systems. Bosch. Ph.D. thesis, Stanford Computer Science, 2001.] • spatial regions: one per glyph – aligned: full glyph, lowest bar component – unaligned: other bar components – task • part-to-whole relationship – scalability • several to one dozen levels for stacked attrib 57

Stacked Bar Chart

http://junkcharts.typepad.com/junk_charts/2013/09/the-inutility-of-stacking-columns.html 58 CS 791/891 - Fall 2015 - Weigle Why not this instead?

http://junkcharts.typepad.com/junk_charts/2013/09/the-inutility-of-stacking-columns.html 59 CS 791/891 - Fall 2015 - Weigle

Idiom: line chart Idiom: line chart 20 • one key, one value 15 • –onedata key, one value 10 –• data2 quant attribs 5 – mark:• 2 pointsquant attribs 0 –• mark:line connection points marks between them – channels • line connection marks between them Year • aligned lengths to express quant value –• channelsseparated and ordered by key attrib into horizontal regions – task• aligned lengths to express quant value • find• separated trend and ordered by key attrib into horizontal regions – task– connection marks emphasize ordering of items along key axis by explicitly showing relationship between one item and the next • find trend – connection marks emphasize ordering of items along key axis by explicitly showing relationship between one item and the next 12

60 Figs 7.12-7.13, p. 151, Few, Now You See It Line Graphs

! When to use: ! When quantitative values change during a continuous period of time

61 CS 791/891 - Fall 2015 - Weigle

p. 171, Tufte, The Visual Display of Quantitative Information Sparklines

! Small, repeated graphics (frequently line graphs)

! jQuerySparklines

62 CS 791/891 - Fall 2015 - Weigle Choosing bar vs line

Choosing bar vs line charts 60 60 • depends on type of key 50 50 attrib 40 40 • depends on type of key attrib 30 30 – bar charts if categorical 20 20 – bar charts if categorical – line charts if ordered 10 10 0 0 – line charts if ordered Female Male Female Male • do not use line charts •fordo categoricalnot use line charts key for 60 60 50 50 attribscategorical key attribs 40 40 30 30 –– violatesviolates expressiveness expressiveness principle 20 20 principle 10 10 • implication of trend so strong that 0 0 10-year-olds 12-year-olds 10-year-olds 12-year-olds • implicationit overrides semantics! of trend so strong that it overrides after [Bars and Lines: A Study of Graphic after [Bars and Lines: A Study of Graphic Communication. semantics!– “The more male a person is, the Communication. Zacks and Tversky. Memory and taller he/she is” CognitionZacks and 27:6 Tversky. (1999), Memory 1073–1079.] and Cognition 27:6 (1999), – “The more male a person is, 1073–1079.] the taller he/she is”

63

13

Idiom:Idiom:Idiom: heatmap heatmap heatmap • two keys, one value • two• two keys, keys, one valueone value – data– data – data • 2 categ• 2 attribscateg (gene, attribs experimental (gene, experimental condition) condition) • 2 categ attribs (gene, experimental condition) • 1 quant• 1 attribquant (expression attrib (expression levels) levels) • 1 quant attrib (expression levels) – marks:– marks: area area – marks:• separate area and align in 2D matrix • separate and align in 2D matrix • separate– indexed and align by in 2 2D categorical matrix attributes – indexed by 2 categorical attributes – channels– indexed by 2 categorical attributes 1 Key 2 Keys Many Keys – channels• color by quant attrib – channels List Matrix Recursive Subdivision • color by– quant (ordered attrib diverging colormap) • color by quant attrib –– (orderedtask diverging colormap) – (ordered diverging colormap) – task • find clusters, outliers – task– scalability • find clusters, outliers • find• clusters,1M items, outliers 100s of categ levels, ~10 quant attrib levels – scalability – scalability • 1M items, 100s of categ levels, ~10 quant attrib levels 14 64 • 1M items, 100s of categ levels, ~10 quant attrib levels 14 Heatmaps

! When to use: ! When you want to display a large quantity of cyclical data

! Black should not be used as a middle value because of its saliency (visual prominence)

! Some people are red-green color blind too

65 CS 791/891 - Fall 2015 - Weigle Few, Now You See It

66 CS 791/891 - Fall 2015 - Weigle Fig 7.22, Few, Now You See It Axis Orientation Rectilinear Parallel Radial

67 16

Idioms:Idioms:Idioms: scatterplot scatterplot scatterplot matrix, matrix, matrix, parallel parallel parallel coordinates coordinates coordinates Scatterplot Matrix Parallel Coordinates • scatterplot• •scatterplotscatterplot matrix (SPLOM)matrix matrix (SPLOM) (SPLOM) Math Physics Dance Drama –rectilinearrectilinear axes, axes, point point mark mark Math – rectilinear– axes, point mark 100 – all possible–– allall pairspossible of axes pairs pairs of axesof axes 90 Physics 80 – scalability 70 – scalability– scalability 60 • one dozen attribs 50 • one dozen• one attribs dozen attribs Dance • dozens to hundreds of items 40 30 • dozens to• dozenshundreds to ofhundreds items of items 20 Drama • parallel coordinates 10 • parallel• –parallelcoordinates parallel coordinates axes, jagged line Math Physics Dance Drama 0 – parallel –axes,representingparallel jagged axes, line jagged itemrepresenting line representing item item Table – rectilinear axes, item as point – rectilinear– rectilinear axes, item axes, as point item as point Math Physics Dance Drama • axis ordering is major challenge • axis ordering• axis is ordering major challenge is major challenge 85 95 70 65 – scalability 90 80 60 50 – scalability– scalability 65 50 90 90 • dozens of attribs 50 40 95 80 • dozens of•• hundreds dozensattribs of attribs of items 40 60 80 90 • hundreds• ofhundreds items of items after [Visualization afterCourse [Visualization Figures. McGuffin, Course 2014. Figures. http://www.michaelmcguffin.com/courses/vis/ McGuffin, 2014. http://www.michaelmcguffin.com/courses/vis/] 17 ] 17 after [Visualization Course Figures. McGuffin, 2014. http://www.michaelmcguffin.com/courses/vis/] 68 20 50 80 100 400 1500 3000 0.0e+00 3.5e+07 10

murder 6 2

Example 80

50 forcible_rape 20 250 Scatterplot Matrix in R robbery 100 0

400 aggravated_assault

! R has built-in scatterplots 100

burglary 1000 400 3000 larceny_theft 1500

! is a multivariate 800 crime motor_vehicle_theft 200

data set 3.5e+07 population

! many columns 0.0e+00 2 6 10 0 100 250 400 1000 200 800 20 50 80 100 400 1500 3000 0.0e+00 3.5e+07 10

! row for each US state murder 6 2 80

50 forcible_rape 20 250 ! robbery Use command to 100 0

show data in all columns 400 aggravated_assault ! plot (crime[,2:9]) 100 burglary 1000 400 3000 larceny_theft

! Add fitted LOESS curve 1500

! plot (crime[,2:9], motor_vehicle_theft 800 200

panel=panel.smooth) 3.5e+07 population

0.0e+00 2 6 10 0 100 250 400 1000 200 800

69 CS 791/891 - Fall 2015 - Weigle Yau, Visualize This, Ch 6 (pp. 180-187)

Task: Correlation • scatterplot matrix – positive correlation • diagonal low-to-high – negative correlation • diagonal high-to-low [A layered grammar of graphics. Wickham. Journ. Computational and Graphical – uncorrelated Statistics 19:1 (2010), 3–28.] • parallel coordinates – positive correlation • parallel line segments – negative correlation • all segments cross at halfway point – uncorrelated • scattered crossings

[Hyperdimensional Using Parallel Coordinates. Wegman. Journ. American Statistical Association 85:411 (1990), 664–675.] 70 Idioms: pie chart,Idioms:Idioms: polar pie pie chart,area chart, polar chart polar area area chart chart • pie chart • pie chartIdioms: pie• chart,pie– area chart marks polar with angle areachannel chart – area marks• pie withchart angle –channel– areaaccuracy: marks angle/area with muchangle less channel accurate than line length – accuracy: angle/area much less accurate than line length – accuracy: –angle/areaarea marks muchwith• polar angle lessarea channel chartaccurate than line length – accuracy: angle/area– area marks much with less length accurate channel than line length • polar– more area direct analogchart to bar charts • polar area chart – area marks with length channel • polar area chart – area marks with length•–data morechannel direct analog to bar charts – area marks with– 1 lengthcateg key channel attrib, 1 quant value attrib – more direct analog to bar charts – more direct •analogtask to bar charts • data– part-to-whole judgements [A layered grammar of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 3–28.] 20 – 1 categ key attrib, 1 quant value attrib • data • data • task – 1 categ key– 1 attrib,categ key 1 quantattrib,– part-to-whole 1value quant attrib value judgments attrib • task • task – part-to-whole judgements [A layered grammar of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 71 – part-to-whole judgements [A layered grammar3–28.] of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 3–28.] 20 [A layered grammar of graphics. Wickham. Journ. Computational and Graphical Statistics 19:1 (2010), 3–28.] 20

Fig 8.4, p. 191-192, Few, Now You See It Part-to-Whole Relationships

! How much wine of different varieties is produced?

72 CS 791/891 - Fall 2015 - Weigle Figs 8.5-8.6, p. 191-192, Few, Now You See It Part-to-Whole Relationships

! How much wine of different varieties is produced?

73 CS 791/891 - Fall 2015 - Weigle

Fig 8.7, p. 191-192, Few, Now You See It Part-to-Whole Relationships ! Bar graphs are much more effective than pie charts for analyzing ranking and part-to-whole relationships

74 CS 791/891 - Fall 2015 - Weigle Idioms: normalized stacked bar chart3/21/2014 bl.ocks.org/mbostock/raw/3886394/ 100%

65 Years and Over 90%

80%

• task 45 to 64 Years 70%

Idioms: normalized stacked60% bar chart Idioms: normalized stacked bar chart3/21/2014 bl.ocks.org/mbostock/raw/3886394/ – part-to-whole judgements 50% 100% 65 Years and Over 90% 25 to 44 Years 40%

80% Idioms: normalized stacked bar chart3/21/2014 bl.ocks.org/mbostock/raw/3886394/ task• task 30% 45 to 64 Years • 70% 100% 18 to 24 Years 65 Years and Over 609%0% 20% 14 to 17 Years – part-to-whole– part-to-whole judgments judgements 508%0% • task 45 to 64 Years 25 to 44 Years 10% 70% 5 to 13 Years 40%

60% 30% Under 5 Years 0% 18 to 24 Years • normalized stacked bar chart– part-to-whole judgements 3/21/2014 50% bl.ocks.org/mbostock/raw/3886208/ UT TX ID AZ NV GA AK MS NMNE CA OK SDCO KS WY NC AR LA IN IL MNDE HI SCMOVA IA TN KY AL WAMDNDOH WI OR NJ MT MI FL NY DC CT PA MAWV RI NHME VT 20% 25 to 44 Years 14 to 17 Years 40%

n 10% 5 to 13 Years o

i 65 Years and Over

t 30% a 35M l 18 to 24 Years u Under 5 Ye4a5rs to 64 Years p 0% • normalized stacked bar chart o 3/21/22001%4 bl.ocks.org/mbostock/raw/3886208/ UT TX ID AZ NV GA AK MS NMNE CA OK SDCO KS WY NC AR LA IN IL MNDE HI SCMOVA IA TN KY AL WAMDNDOH WI OR NJ MT MI FL NY DC CT PA MAWV RI NHME VT – stacked bar chart, normalized to full vert height P 14 to 17 Years 25 to 44 Years 10% 5 to 13 Years n 18 to 24 Years o

i 65 Years and Over t a 30M 35M l u Under 5 Years 45 to 64 Years

p 14 to 17 Years 0% 3/21/2014 o bl.ocks.org/mbostock/raw/3886208/

• normalized stacked bar chart P UT TX ID AZ NV GA AK MS NMNE CA OK SDCO KS WY NC AR LA IN IL MNDE HI SCMOVA IA TN KY AL WAMDNDOH WI OR NJ MT MI FL NY DC CT PA MAWV RI NHME VT 25 to 44 Years – stacked bar chart, normalized to full vert height 5 to 13 Years 18 to 24 Years

30M n o i 65 Years and Over 14 to 17 Years t Under 5 Years a 35M l u 45 to 64 Years 25M p 5 to 13 Years • normalized stacked bar chart o P 25 to 44 Years – stacked bar chart, normalized to full vert height Under 5 Years – single stacked bar equivalent to full pie 18 to 24 Years 25M 30M – single stacked bar equivalent to full pie 14 to 17 Years 5 to 13 Years 20M Under 5 Years – stacked• high bar– informationsingle chart, stacked normalized density: bar equivalent requires to full tonarrow vertfull pie rectangleheight 202M5M 20M • high information density: requires narrow rectangle 15M • high information density: requires narrow15 Mrectangle 15M – single stacked bar equivalent to full pie 10M

10M 10M 5.0M • high information density: requires narrow rectangle 5.0M 0.0 • pie chart CA TX NY FL IL PA OH MI GA3N/2C1/N2J01V4A WA AZ MA IN TN MO MD WI MN CO AL SC LA KY OR OK CT IAblM.oScAksR.oKrgS/mUbToNstVocNkM/rWawV/N38E86ID39M4E/ NH HI RI MT DE SD AK ND VT DC WY 5.0M 0.0 • pie chart CA TX NY FL IL PA OH MI GA3N/2C1/N2J01V4A WA AZ MA IN TN MO MD WI MN CO AL SC LA KY OR OK CT IAblM.oScAksR.oKrgS/mUbToNstVocNkM/rWawV/N38E86ID39M4E/ NH HI RI MT DE SD AK ND VT DC WY 100% 3/21/2014 bl.ocks.org/mbostock/raw/3887235/ 100% 3/21/2014 bl.ocks.org/mbostock/raw/3887235/ 65 Years and Over 90% 65 Years and Over – information density: requires large circle 0.0 90% – information density: requires large circle 80% • pie chart CA TX NY FL IL PA OH MI GA3N/2C1/N2J01V4A WA AZ MA IN TN MO MD WI MN CO AL SC LA KY OR OK CT IAblM.oScAksR.oKrgS/mUbToNstVocNkM/rWawV/N38E86ID39M4E/ NH HI RI MT DE SD AK ND VT DC WY 80%

≥65 <5 45 to 64 Years 70% ≥65 70% <5 45 to 64 Years 100% 45-64 5-13 3/21/2014 b45l-6.4ocks.org/mbostoc5k-1/3raw/3887235/ 60%60% 65 Years and Over 14-17 14-17 90% 50%50%

18-24 25 to 44 Years 40% 18-24 25 to 44 Years – information density: requires large circle 40% http://bl.ocks.org/mbostock/3887235, 80% http://bl.ocks.org/mbostock/3887235, http://bl.ocks.org/mbostock/raw/3886394/ 1/1 http://bl.ocks.org/mbostock/raw/388639430/% 25-44 1/1 30% 25-44 18 to 24 Years 18 to 24 Years 20% ≥65 pie chart http://bl.ocks.org/mbostock/3886208, <5 14 to 17 Years 45 to 64 Years • 70% http://bl.ocks.org/mbostock/3886208, 20% 14 to 17 Years 10% 5 21to 13 Years 45-64 5-13 5 to 13 Years http://bl.ocks.org/mbostock/3886394. 10% Under 5 Years 21 60% 0% UT TX ID AZ NV GA AK MS NMNE CA OK SDCO KS WY NC AR LA IN IL MNDE HI SCMOVA IA TN KY AL WAMDNDOH WI OR NJ MT MI FL NY DC CT PA MAWV RI NHME VT http://bl.ocks.org/mbostock/3886394. Under 5 Years 0% information density: requires large circle UT TX ID AZ NV GA AK MS NMNE CA OK SDCO KS WY NC AR LA IN IL MNDE HI SCMOVA IA TN KY AL1W4-A1M7DNDOH WI OR NJ MT MI FL NY DC CT PA MAWV RI NHME VT – 50%

18-24 25 to 44 Years http://bl.ocks.org/mbostock/3887235, 40% http://bl.ocks.org/mbostock/raw/3886394/ 1/1 30% 25-44 18 to 24 Years http://bl.ocks.org/mbostock/3886208, 20% 14 to 17 Years

10% 5 21to 13 Years

http://bl.ocks.org/mbostock/3886394. Under 5 Years 0% http://bUl.TocTkXs.orIDg/mAbZosNtoVckG/rAawA/K38M86S2N08M/ NE CA OK SDCO KS WY NC AR LA IN IL MNDE HI SCMOVA IA TN KY AL WAMDNDOH WI OR NJ MT MI FL N1/Y1 DC CT PA MAWV RI NHME VT

http://bl.ocks.org/mbostock/3887235, http://bl.ocks.org/mbostock/raw/3886208/ 1/1

http://bl.ocks.org/mbostock/raw/3887235/ 1/1 http://bl.ocks.org/mbostock/3886208,

http://bl.ocks.org/mbostock/raw/3887235/ 1/1 http://bl.ocks.org/mbostock/3886394. 75 http://bl.ocks.org/mbostock/raw/3886394/ 1/1

http://bl.ocks.org/mbostock/raw/3886394/ 1/1

http://bl.ocks.org/mbostock/raw/3886208/ 1/1

http://bl.ocks.org/mbostock/raw/3887235/ 1/1 IdiomIdiom designdesign choices: choices: Part Part1 1 Encode http://bl.ocks.org/mbostock/raw/3886394/ 1/1

Arrange Express Separate from categorical and ordered attributes Color What? Hue Saturation Luminance Order Align Why? Size, Angle, Curvature, ... How?

Use Shape

Motion Direction, Rate, Frequency, ...

76 3 Idiom design choices: Part 2

77

Manipulate

78 Facet

79

Juxtapose and coordinate views

80 Idiom: Linked highlighting System: EDV • see how regions contiguous in one view are distributed within another – powerful and pervasive interaction idiom

• encoding: different – multiform • data: all shared

[Visual Exploration of Large Structured Datasets. Wills. Proc. New Techniques and Trends in Statistics (NTTS), pp. 237–246. IOS Press, 1995.]

81

Idiom: Small multiples System: Cerebral • encoding: same • data: none shared – different attributes for node colors – (same network layout) • navigation: shared

[Cerebral: Visualizing Multiple Experimental Conditions on a Graph with Biological Context. Barsky, Munzner, Gardy, and Kincaid. IEEE Trans. Visualization and (Proc. InfoVis 2008) 14:6 (2008), 1253–1260.] 82 Coordinate views: Design choice interaction

83

Juxtapose design choices • design choices – view count • few vs many – how many is too many? open research question – view visibility • always side by side vs temporary popups – view arrangement • user managed vs system arranges/aligns • why juxtapose views? – benefits: eyes vs memory • lower cognitive load to move eyes between 2 views than remembering previous state with 1 – costs: display area • 2 views side by side each have only half the area of 1 view

84 System: Improvise

• investigate power of multiple views – pushing limits on view count, interaction complexity – reorderable lists • easy lookup • useful when linked to other encodings

[Building Highly-Coordinated Visualizations In Improvise. Weaver. Proc. IEEE Symp. Information Visualization (InfoVis), pp. 159–166, 2004.] 85

PartitionPartition intointo views • how• how to to divide divide data between between views Partition Into Views views– encodes association between items – encodesusing spatial association proximity between items– major using implications spatial for proximity what patterns – majorare visible implications for what patterns are visible – split according to attributes – split according to attributes • design• design choiceschoices – how– how many many splitssplits • all• all the the way way down: one one mark mark per perregion? region?• stop earlier, for more complex structure • stopwithin earlier, region? for more complex structure within region? – order– order in in which which attribsattribs usedused to to split split– how many views 25 – how many views 86 ViewsPartition and into glyphs views • view• how to divide data between views Partition Into Views – contiguous– encodes association region in between which items visuallyusing spatial encoded proximity data is shown on– major the display implications for what patterns • glyphare visible – object– split accordingwith internal to attributes structure that arises from multiple marks • no• design strict choices dividing line – view:– how big/detailedmany splits – glyph:small/iconic• all the way down: one mark per region? • stop earlier, for more complex structure within region? – order in which attribs used to split – how many views 25 87

Partitioning: List alignment Partitioning: List alignment •Partitioning:single bar chart List with alignment grouped bars • small-multiple bar charts • single–• splitsingle barby state chartbar into chart with regions groupedwith grouped bars bars •• small-multiplesmall-multiple– split by age barinto bar chartsregions charts – split– split by state by intostate regions into regions –– splitsplit byby age age into into regions regions • complex• complex glyph withinglyph eachwithin region each showing region all ages • one• one chart chart per per region region – compare:• complexshowing easyglyph within all ages each state, region hard showing across all ages ages – compare:–• onecompare: chart per easyeasy region within within age, age, harder harder – compare: easy within state, hard across states – compare:across easy ages within state, hard across ages – compare:across easystates within age, harder

11across states 5 11.0 65 Years and Over 45 to 64 Years 0 10.0 25 to 44 Years 11 18 to 24 Years 5 9.0 14 to 17 Years 0 5 to 13 Years 11 8.0 Under 5 Years 5 7.0 0 11 6.0 5 0 5.0 11 5 4.0 0 3.0 11 5 2.0 0 11 1.0 5 0 0.0 CA TK NY FL IL PA 27 CA TK NY FL IL PA 27 88 Trellis Display

! Multiple graphs that typically vary on one variable ! Arrange in order of magnitude, not alphabetically or arbitrarily

89 CS 791/891 - Fall 2015 - Weigle Few, Now You See It

Arranged alphabetically by division

90 CS 791/891 - Fall 2015 - Weigle Fig 5.11, Few, Now You See It Arranged by magnitude

91 CS 791/891 - Fall 2015 - Weigle Fig 5.12, Few, Now You See It

Superimpose layers • layer: set of objects spread out over region – each set is visually distinguishable group – extent: whole view • design choices – how many layers? – how are layers distinguished? – small static set or dynamic from many possible? – how partitioned? • heavyweight with attribs vs lightweight with selection • distinguishable layers – encode with different, nonoverlapping channels • two layers achieveable, three with careful design

92 Static visual layering • foreground layer: roads – hue, size distinguishing main from minor – high luminance contrast from background • background layer: regions – desaturated colors for water, parks, land areas • user can selectively focus attention • “get it right in black and white” – check luminance contrast with greyscale view [Get it right in black and white. Stone. 2010. http://www.stonesc.com/wordpress/2010/03/get-it-right-in-black-and- white] 93

Superimposing limits • few layers, but many lines – up to a few dozen – but not hundreds • superimpose vs juxtapose: empirical study – superimposed for local visual, multiple for global – same screen space for all multiples, single superimposed – tasks • local: maximum, global: slope, discrimination

[Graphical Perception of Multiple Time Series. Javed, McDonnel, and Elmqvist. IEEE Transactions on Visualization and Computer Graphics (Proc. IEEE InfoVis 2010) 16:6 (2010), 927–934.]

94 Reduce items and attributes • reduce/increase: inverses • filter – pro: straightforward and intuitive • to understand and compute – con: out of sight, out of mind • aggregation – pro: inform about whole set – con: difficult to avoid losing signal • not mutually exclusive – combine filter, aggregate – combine reduce, change, facet

95

Idiom: scented widgets Idiom: scented widgets • augment widgets for filtering to show information scent • augment– cues to show widgets whether valuefor filtering in drilling downto show further information vs looking elsewhere scent – cues to show whether value in drilling down further vs looking • concise,elsewhere in part of screen normally considered control panel • concise, in part of screen normally considered control panel

[Scented Widgets: Improving Navigation Cues with Embedded Visualizations. Willett, Heer, and Agrawala. IEEE Trans. Visualization and Computer Graphics (Proc. InfoVis 2007) 13:6 (2007), 1129–1136.] [Scented Widgets: Improving Navigation Cues with Embedded Visualizations. Willett, Heer, and Agrawala. IEEE Trans. Visualization and Computer Graphics (Proc. InfoVis 2007) 13:6 (2007), 1129– 1136.] 40

96 Idiom: histogram Idiom: histogram 20 • static item aggregation 15 • •statictask: finditem distribution aggregation 10 • •task:data: findtable distribution 5 0 • •data:derived table data – new table: keys are bins, values are counts • derived data Weight Class (lbs) • bin size crucial – new table: keys are bins, values are counts – pattern can change dramatically depending on discretization • bin size crucial – opportunity for interaction: control bin size on the fly – pattern can change dramatically depending on discretization – opportunity for interaction: control bin size on the fly

42

97

Example Histogram in R Histogram of birth$X2008

! R has a built-in 60

histogram 50 40 30 Frequency

! X2008 is a column 20 in the birth dataset 10 0 Histogram of birth$X2008 10 20 30 40 50

birth$X2008 35

! Use hist command 30 ! hist(birth$X2008) 25 20 Frequency 15

! Add more bins 10 ! hist(birth$X2008, 5 0

breaks=20) 10 20 30 40 50

birth$X2008

98 CS 791/891 - Fall 2015 - Weigle Yau, Visualize This, Ch 6 (pp. 203-208) Example Density (Frequency Polygon) in R

! Use birth rate data density.default(x = birth2008)

! Clean data for 2008 (remove values with NA) 0.04 ! data can't have missing values for density 0.03 computation Density

0.02 ! Use density command to

compute stats 0.01 ! d2008 <- density(birth2008) 0.00

0 10 20 30 40 50 60

! Use plot command N = 219 Bandwidth = 3.168 ! plot (d2008)

99 CS 791/891 - Fall 2015 - Weigle Yau, Visualize This, Ch 6 (pp. 208-213)

Figs 10.25-10.26, p. 227, Few, Now You See It Strip Plot

! Display each individual value in a distribution, laid out on a single line

100 CS 791/891 - Fall 2015 - Weigle pod, and the rug plot looks like the seeds within. Kampstra (2008) also suggests a way of comparing two

groups more easily: use the left and right sides of the bean to display different distributions. A related idea

is the raindrop plot (Barrowman and Myers, 2003), but its focus is on the display of error distributions from

complex models.

Figure 4 demonstrates these density boxplots applied to 100 numbers drawn from each of four distribu-

tions with mean 0 and standard deviation 1: a standard normal, a skew-right distribution (Johnson distri-

bution with skewness 2.2 and kurtosis 13), a leptikurtic distribution (Johnson distribution with skewness 0

and kurtosis 20) and a bimodal distribution (two normals with mean -0.95 and 0.95 and standard devia-

tion 0.31). Richer displays of density make it much easier to see important variations in the distribution: Idiom: boxplot multi-modality is particularly important, and yet completely invisible with the boxplot. • Idiom:static item boxplot aggregation ! !

! ! 4 4

! • task: find distribution ! ! 4 ! • static item aggregation 4

! !

• •data: task: table find distribution ! !

2 2 !

! 2 • •derived data: datatable 2 0 0 •– derived5 quant attribs data 0 0

! – 5 quant attribs ! ! • median: central line ! !

! 2

! 2 ! ! • median: central line ! !

2 2 !

• lower and upper quartile: boxes ! ! ! ! • lower and upper quartile: boxes ! !

! ! 4

• lower upper fences: whiskers ! 4

• lower upper fences: whiskers ! – values beyond which items are outliers – values beyond which items are outliers n s k mm n s k mm n s k mm n s k mm – outliers– outliers beyond beyond fence fence cutoffs cutoffs explicitly explicitly shown shown Figure 4: From left to right: box plot, vase plot, violin plot and bean plot. Within each plot, the distributions from left to [40 years of boxplots. Wickham and Stryjewski. 2012. had.co.nz] right are: standard normal (n), right-skewed (s), leptikurtic (k), and bimodal (mm). A normal kernel and bandwidth of [40 years of boxplots. Wickham and Stryjewski. 2012. had.co.nz] 43

0.2 are used in all plots for all groups.101

A more sophisticated display is the sectioned density plot (Cohen and Cohen, 2006), which uses both

Box Plots colour and space to stack a density estimate into a smaller area, hopefully without losing any information

! bottom of box - 25th percentile(not (1 formallyst quartile) verified with a perceptual study). The sectioned density plot is similar in spirit to horizon ! top of box - 75th percentile (3rd quartile) graphs for time series (Reijner, 2008), which have been found to be just as readable as regular line graphs ! band inside box - mean despite taking up much less space (Heer et al., 2009). The density strips of Jackson (2008) provide a similar ! When to use: compact display that uses colour instead of width to display density. These methods are shown in Figure 5. ! You want to show how values are distributed across a range and how that distribution changes over time ! Can also be used to compare distributions without the time dimension 6

102 CS 791/891 - Fall 2015 - Weigle Few, Now You See It 103 CS 791/891 - Fall 2015 - Weigle Fig 7.23, Few, Now You See It

Example Box Plots in R

! R has built-in box plots 100

! boxplot() function 50 ! whiskers - 1.5 IQR (inter-quartile

range) 20 ! dots - high or low values outside 1.5 IQR 10 5 ! Example using built-in data OrchardSprays 2

A B C D E F G H > OrchardSprays decrease rowpos colpos treatment 1 57 1 1 D 2 95 2 1 E

boxplot(decrease ~ treatment, data = OrchardSprays, log = "y", col = "bisque")

104 CS 791/891 - Fall 2015 - Weigle A Tour Through the Visualization Zoo

Heer, Bostock, Ogievetsky, "A Tour Through the Visualization Zoo", ACM Queue, 2010

105 CS 791/891 - Fall 2015 - Weigle

Is there a perfect vis for a particular data type? ! Task?

! User?

! "It's not possible to decide for a given visualizations whether it is optimal, or perfect", the underlying mechanics of the question are still useful to ponder."

! "The perfect is the enemy of the good, and it is also the enemy of the practical, useful solution."

http://eagereyes.org/blog/2013/perfect-visualization

106 CS 791/891 - Fall 2015 - Weigle Outline

! Motivation -- why vis?

! Tools

! What, Why, How Framework

! How ! marks and channels ! chart types

107 CS 791/891 - Fall 2015 - Weigle