Visualizing Information

139 Pages·2008·3.677 MB·English

by  Riccardo Mazza

Riccardo Mazza
Introduction to Information
Visualization
123
RiccardoMazza
UniversityofLugano
Switzerland
ISBN:978-1-84800-218-0 e-ISBN:978-1-84800-219-7
DOI:10.1007/978-1-84800-219-7
BritishLibraryCataloguinginPublicationData
AcataloguerecordforthisbookisavailablefromtheBritishLibrary
LibraryofCongressControlNumber:2008942431
(cid:2)c Springer-VerlagLondonLimited2009
Apart from any fair dealing for the purposes of research or private study, or criticism or review, as
permittedundertheCopyright,DesignsandPatentsAct1988,thispublicationmayonlybereproduced,
storedortransmitted,inanyformorbyanymeans,withthepriorpermissioninwritingofthepublish-
ers,orinthecaseofreprographicreproductioninaccordancewiththetermsoflicencesissuedbythe
CopyrightLicensingAgency.Enquiriesconcerningreproductionoutsidethosetermsshouldbesentto
thepublishers.
Theuseofregisterednames,trademarks,etc.,inthispublicationdoesnotimply,evenintheabsenceofa
specificstatement,thatsuchnamesareexemptfromtherelevantlawsandregulationsandthereforefree
forgeneraluse.
Thepublishermakesnorepresentation,expressorimplied,withregardtotheaccuracyoftheinformation
containedinthisbookandcannotacceptanylegalresponsibilityorliabilityforanyerrorsoromissions
thatmaybemade.
Printedonacid-freepaper
SpringerScience+BusinessMedia
springer.com
Contents
1 IntroductiontoVisualRepresentations ........................... 1
1.1 Presentation ............................................... 4
1.2 ExplorativeAnalysis ....................................... 5
1.3 ConfirmativeAnalysis ...................................... 6
1.4 InformationVisualization.................................... 7
1.5 FromDatatoWisdom....................................... 8
1.6 MentalModels............................................. 10
1.7 ScientificVisualization...................................... 11
1.8 CriteriaforGoodVisualRepresentations....................... 12
1.8.1 GraphicalExcellence................................. 13
1.8.2 GraphicalIntegrity................................... 13
1.8.3 MaximizetheData–InkRatio.......................... 14
1.8.4 Aesthetics .......................................... 14
1.9 Conclusion................................................ 15
2 CreatingVisualRepresentations ................................ 17
2.1 AReferenceModel ........................................ 17
2.1.1 PreprocessingandDataTransformations................. 18
2.1.2 VisualMapping ..................................... 20
2.1.3 Views.............................................. 23
2.2 DesigningaVisualApplication .............................. 24
2.3 VisualRepresentationofLinearData ......................... 26
2.4 2Dvs.3D................................................. 30
2.5 Conclusion................................................ 31
3 Perception..................................................... 33
3.1 Memory .................................................. 33
3.2 PreattentiveProperties ...................................... 35
3.2.1 Color .............................................. 35
3.2.2 Form .............................................. 35
3.2.3 SpatialPosition...................................... 35
3.2.4 Movement.......................................... 38
3.3 MappingDatatoPreattentiveAttributes........................ 38
3.4 PostattentiveProcessing ..................................... 39
3.5 GestaltPrinciples........................................... 41
3.5.1 FigureandGround................................... 42
3.5.2 Proximity........................................... 42
3.5.3 Similarity........................................... 42
3.5.4 Closure ............................................ 43
3.5.5 Continuity .......................................... 43
3.5.6 OtherPrinciples ..................................... 44
3.6 Conclusion................................................ 44
4 MultivariateAnalysis .......................................... 45
4.1 TheProblemofMultivariateVisualization...................... 46
4.2 GeometricTechniques ...................................... 47
4.2.1 ParallelCoordinates.................................. 47
4.2.2 ScatterplotMatrix.................................... 50
4.2.3 TableLens .......................................... 52
4.2.4 ParallelSets......................................... 52
4.3 IconTechniques............................................ 55
4.3.1 StarPlots........................................... 55
4.3.2 ChernoffFaces ...................................... 57
4.4 Pixel-OrientedTechniques................................... 59
4.5 Conclusion................................................ 62
5 NetworksandHierarchies ...................................... 63
5.1 NetworkData.............................................. 64
5.1.1 ConceptMapsandMindMaps......................... 65
5.1.2 ComplexNetworkData............................... 66
5.1.3 GeographicRepresentations ........................... 72
5.1.4 TransportNetworks .................................. 73
5.1.5 3DGraphs.......................................... 74
5.2 HierarchicalData .......................................... 75
5.2.1 FileSystem ......................................... 77
5.2.2 RepresentingEvolutionaryDatawithTrees .............. 79
5.2.3 ConeTree .......................................... 81
5.2.4 BotanicalTree....................................... 81
5.2.5 Treemap............................................ 83
5.3 Conclusion................................................ 88
6 WorldWideWeb .............................................. 91
6.1 WebsiteMaps ............................................. 91
6.2 WebsiteLogData .......................................... 93
6.3 VisualRepresentationofSearchEngineResults ................. 98
6.3.1 Clustering ..........................................100
6.4 AnalysisofInteractionsinBlogs..............................102
6.5 Conclusion................................................103
7 Interactions ...................................................105
7.1 TheProblemofInformationOverload .........................105
7.2 TypesofInteractiveVisualRepresentations.....................106
7.3 ManipulableRepresentations.................................107
7.3.1 Scrolling ...........................................107
7.3.2 Overview+Details ..................................107
7.3.3 Focus+Context .....................................110
7.4 TransformableRepresentations ...............................116
7.4.1 FilteringInputData ..................................116
7.4.2 DataReordering .....................................117
7.4.3 DynamicQueries ....................................117
7.4.4 MagicLens .........................................119
7.4.5 AttributeExplorer ...................................121
7.5 Conclusion................................................122
8 Evaluations ...................................................125
8.1 Human–ComputerInteraction ................................125
8.2 EvaluationCriteria .........................................126
8.3 EvaluatingVisualRepresentations ............................127
8.3.1 AnalyticMethods....................................128
8.3.2 EmpiricalMethods...................................129
8.4 Conclusion................................................132
References.........................................................133
Index .............................................................137
Chapter 1
Introduction to Visual Representations
Let’sstopforamomentandconsiderjusthowmuchinformationwehavetotakein
everydayas partofour routineactivities. E-mailsarriveon ourcomputers,credit
card statementsarrive fromthe bankeverymonth, and last-minuteholidayoffers,
stockmarketindexvariations,andadvertisingleafletsfillthemailbox.Nottomen-
tion work. Perhaps you work in a large department store and have to decide the
discountpoliciestobeappliedtosaleitems:Whichitemsshouldweputonsalein
thecomingmonths?Summerisarriving—shouldweperhapsputthebeachumbrel-
las on sale? What percentagediscountshouldwe apply?How did the sales of the
previousmonth’spromotionalitemsgo?
Inallofthesesituations,thecommonrecurringthemeistheenormousquantity
of information that we have to deal with on a daily basis. Each of the previously
described situations almost always involves making a decision: Which e-mail or
advertisingflyercanwethrowoutbecauseitdoesn’tinterestus?Howmuchdidwe
chargetothecreditcardlastmonth?Willweperhapsneedtolimitourspendingin
the future? Where can we spend the next holiday withoutit costing us a fortune?
Would it be worthwhileto investour savingsin a particularstock? Whatdiscount
canweputonthebeachumbrellasinthecomingmonths?
Perhaps we haven’teven realized, but in the last decade, the quantity of infor-
mationthatweallhavetoprocesshasincreasedenormously.Theglobalizationof
economyandcommunication,butabovealltherapidadvancesintechnology(and
not only communication and information technology), have brought us in, recent
years,towhatsomenotedscholarsdefineasinformationpollution.Anyway,ifwe
really think about it, what we are witnessing in reality is not an explosion of in-
formation, but rather an explosion of data, which we are continuously pressed to
observe,process,anddevelop,forourfamilyorworkactivities.Weareinformedby
thedatathatwecontinuallyreceivefromnumeroussources.Theinformation,very
valuable and important for our lives, is built and elaborated on starting from this
continuousandconstantinfluxofdatathatwearepassivelyoractivelysubjectedto.
Therefore,weneedeffectivemethodsthatallowustogothroughthisinformation
and,forexample,helpusmakedecisions.
R.Mazza,IntroductiontoInformationVisualization, 1
DOI:10.1007/978-1-84800-219-7 1,(cid:2)c Springer-VerlagLondonLimited2009
2 1 IntroductiontoVisualRepresentations
Fig. 1.1 Road map for theLugano–Pisa route, provided ina textual version (left) and a visual
version(right).Imagefromhttp://www.viamichelin.com;reproducedwiththepermissionofVia-
Michelin.
Therearenumeroussituationsinwhichweusevisualrepresentationstounder-
stand the variousdata. Thiscould involveanythingfrom last week’sstock market
trendstoatravelitineraryoreventheweatherforecastforvariousgeographicalar-
eas. Thanks to our visual perception ability, a visual representation is often more
effectivethanwrittentext.
Let’stake,forinstance,thecaseofapersonwhohastotravelbycarfromLugano
to Pisa and needs to find out which route to take. It is possible to represent this
informationinatextualformbyproviding,forexample,ameticulousdescriptionof
theroadstofollowandthejunctionstotake.Itis,however,alsopossibletorepresent
thisinformationin avisualform,througha mapthatvisuallyhighlightsthe entire
routetofollow.A routegeneratedbyaverypopularwebsiteisrepresentedinFig.
1.1.
The website in Fig. 1.1 providesa very useful service. We can set a departure
pointandadestination,andthewebsitewillindicatetheroutetofollow.Amongthe
various configurable options, we can request an itinerary that favors the highway
orthetoll-freeroads.Thewebsitecreatesthebestroutepossible,accordingtoour
requirements.Theroute,aswecansee inFig.1.1,ispresentedintwoforms:One
isatextualtablethatreportsthedistances,thenamesoftheroadstofollow,andthe
junctionstonote,andtheotherisavisualversionintheformofaroadmap.
1 IntroductiontoVisualRepresentations 3
Thewebsiteprovidestwocomplementaryversionsthatcanbeusedfordifferent
purposes.Forexample,atruckdrivertransportinggoodswillwanttoknowexactly
which roads to take and their relative distances; in this case, the textual version
canbeveryuseful.Thereare,however,someaspectsthatcanbeinterestingwhen
weplananitineraryforarecreationaljourney,suchasthepossibilityoffindingan
alternativerouteorplacesclosetotheroutethatmightbeofinteresttothetourist.
Althoughuselessforthetruckdriver,theseaspectscouldindeedbeindispensablefor
afamilywishingtoprogramtheroutefortheirnextholidayandcanbeeffectively
revealedthroughtheuseofthevisualversionoftheroute.
Thevisualversionhastheadvantageofusingsomegraphicalpropertiesthatare
very quickly and efficiently processed by visual perception. The visual attributes
like color,size, proximity,andmovementare immediatelytaken in andprocessed
bytheperceptualabilityofvision,evenbeforethecomplexcognitiveprocessesof
thehumanmindcomeintoplay.
Let’s clarifythisconceptwith anexample.Figure1.2showsa sequenceof nu-
merical data and a visual representation,constructedby horizontallines of length
proportionaltothevaluesontheleftthattheyrepresent.
320
260
380
280
420
400
Fig.1.2 Mappingnumericalvaluestothelengthsofbars.
Let’ssupposethatwehavetodeterminethemaximumandminimumnumerical
values indicated on the left. If we didn’thave the lines at our disposal, we would
havetoperformthefollowingprocedure:Readeachofthenumericalvalues,keep-
ing in mind the extreme values (the maximum and the minimum) that we come
acrosswhilereadingthem,rightthroughtotheend.Inonesensethisisacognitive
exercise,sinceitisnecessarytocomparethepairsofnumericalvalueseachtimeto
decideifonevalueishigherorlowerthantheother.
We’llrepeatthesameexercise,thistimewiththeaidofthelinesontheright.The
length of the lines shows us at a glance the maximum and minimum values. This
4 1 IntroductiontoVisualRepresentations
information is processed by our visual perception, which immediately recognizes
thelengthsofthelinesandarrangestheminrelationshiptothevaluesrepresented.
Sincehumansperceivevisualattributesverywell,liketheextensionofthelines
in the previouscase, we canrepresenta greatdealofdifferentdata by“mapping”
themtodifferentvisualattributes.Forinstance,wecouldrepresentthelinesofthe
previousfigurewith differentcolors,ordifferentwidths, to codifyfurtherdata.In
thiscase,thevisualrepresentations,ifwellconstructed,canbeusefulnotonlyfor
perceivinginformationmorequicklybutalsoforprocessingseveralitemsofinfor-
mationatthesametime.Let’snotforgetthatthehumanbrainisa“machine”that
constantlyprocessesahugeamountofdataandinformationsimultaneously.Inthis
waywecaneasilysingleout,inoneormorecollectionsofdata,themaximumand
minimumvalues,theexistenceofrelationshipsbetweenthedata,grouping,trends,
gaps,orinterestingvalues.Asaresult,thevisualrepresentationsallowustounder-
standcomplexsystems,makedecisions,andfindinformationthatotherwisemight
remainhiddeninthedata.
1.1 Presentation
Whenwewanttocommunicateanidea,wesometimesuseapicture.Itcouldbea
sketchonpaper,adrawingonablackboard,orimagesprojectedonaslideortrans-
parency.Thevisualrepresentationshelpustoillustrate conceptsthat,if expressed
verbally,wewouldfinddifficulttoexplainclearlytoalistener.Justimaginetrying
to explainto someoneoverthe telephonehow to fix a bathroomfaucet. When we
havedatawithwhichweneedtoillustrateconcepts,ideas,andpropertiesintrinsicto
thatdata,theuseofvisualrepresentationoffersusavalidcommunicationtool.The
difficultpartisindefiningtherepresentationsthateffectivelyachievetheirgoal.Ed-
wardTufte,oneofthemajorcontemporaryscholarsofthisdisciplineandProfessor
EmeritusofPoliticalScience,Statistics, andComputerScienceatYaleUniversity,
maintainsthat“excellenceinstatisticalgraphicsconsistsofcomplexideascommu-
nicatedwithclarity,precision,andefficiency”[58].Itis necessaryfora pictureto
give the reader as much data as can be processed quickly, using as little space as
possible.
Let’slookatthevisualrepresentationillustratedinFig.1.3.Itdealswithamap
createdbyCharlesJosephMinard,a Frenchengineer,in 1869.The mapwascon-
ceived to illustrate the number of losses suffered by Napoleon’s army during the
disastrousmarchtowardMoscowin1812.Thethickbandshowstheroutetakenby
thetroupes,fromthePolishbordertoMoscow,andthewidthofthistrackrepresents
the numberof soldiers presentat each pointof the journey.The numberof losses
suffered by the army is evident at a glance. Of the 422,000 soldiers who set off
fromthePolishborder,only100,000arrivedinMoscow.Napoleon’sretreatduring
the freezing Russian winter is representedby the dark line, linked to a graph that
reportsthe harsh temperaturesthat further decimated the already-exhaustedarmy.
Somerivers,inwhichnumeroussoldierslosttheirlivesattemptingtocross,arealso
1.2 ExplorativeAnalysis 5
Fig.1.3 VisualrepresentationofthemarchofNapoleon’sarmyintheRussiancampaignof1812,
producedbyCharlesJ.Minard.
indicated.Thisvisualisasuperbexampleoftheconceptofexcellenceexpressedby
Tufte,who,notwithoutgoodreason,defineditas“thebeststatisticalgraphicever
drawn”[58].
1.2 ExplorativeAnalysis
The explorative analysis of data is one of the applications that benefits the most
fromvisualrepresentationsandtheabilityofanalysisbyvisualperceptionandthe
humancognitivesystem. This has been used for yearsto identify properties,rela-
tionships, regularities, or patterns. Jacques Bertin (a French cartographer who, as
earlyas1967,wroteaworkdefiningthebasicelementsofeveryvisualrepresenta-
tion)definesitas“thevisualmeansofresolvinglogicalproblems”[5].
We’ll illustrate the conceptwith an example. Figure 1.4 displays some statisti-
caldata oncancer-relatedmortalityamongmenin the United States in the period
from1970to 1994.In the picture,the countiesare represented(3,055in total) by
a colorscale rangingfromblueto red,accordingto thepercentageofcasesfound
ineachcounty.Thankstothecolor,wecansingleoutthegeographicalareaswith
an average (white), below-average(blue shades), and above-average(red shades),
numberofcases.Itisnoticeablehowaboveaverage-casesarepredominantlyfound
inthecountiesalongtheEastCoastandinthesoutheastoftheUnitedStates.The
AmericanNationalCancerInstituteproducedthisandmanyotherimageswiththe
aimofidentifyingpossiblecausesfortheonsetoftumors.Infact, itis bynowal-
mostcertainthatmostcasesofcancerareassociatedinsomewaywithlifestylesthat