overcomefusioncharts

fusioncharts  时间:2021-05-20  阅读:()
Non-spatialVisualisation8KarelMackAbstractAnenormousamountofvariousdataispro-ducedeveryday.
Withproperdatavisualisation,aninformationhiddeninthedatacanbeeasilyandquicklyrevealed.
Itisnecessarytocreateacommunicationchannelthatcouldquicklyandefcientlytransfertheinformationfromthedatatotheuser.
Byusingvisualelementslikecharts,graphs,andmaps,datavisualisationisanaccessiblewaytoseeandunderstandtrends,outliers,andpatternsindata.
Thischapteroffersanoverviewofrele-vantdatavisualisationsdividedintothematiccategoriesandsupportedbyexamples.
KeywordsVisualisation·Data·Chart·InformationIntheworldtoday,weencounterenormousamountsofdataeveryday.
Toconvertdataintousefulinformation,datamustbepresentedtotheuserinawaythatallowsinterpreting,analysingandapplyingthegainedinformation(Yau2011).
Itisnecessarytocreateacommunicationchannelthatcouldquicklyandefcientlytransfertheinformationfromthedatatotheuser–thiscanbedonewithdatavisualisation.
TableauSoft-ware,acompanyofferingasoftwareplatformforinteractivedatapresentation,brieyandcom-prehensivelytalksaboutdatavisualisation:"Datavisualisationreferstothegraphicalrepresentationofinformationanddata.
Byusingvisualelementslikecharts,graphs,andmaps,datavisualisationisanaccessiblewaytoseeandunderstandtrends,outliers,andpatternsindata"(TableauSoftware2018).
Accordingto(Friedman2008,p.
1)the"maingoalofdatavisualisationistocommuni-cateinformationclearlyandeffectivelythroughgraphicalmeans.
Toconveyideaseffectively,bothaestheticformandfunctionalityneedtogohandinhand,providinginsightsintoarathersparseandcomplexdatasetbycommunicatingitskey-aspectsmoreintuitively".
Visualisationisanimportantstepinthewholeprocessofdataanalysis.
LegendarystatisticianJohnTukeyoftenmentionsvisualisationinthecontextofusingvisualisationtondmeaninginthedata(Tukey1977).
Despitehisstatisticalfocus,hebelievedthatagraphicpresentationofinformationplaysanimmenserole.
Apropervisualisationbasedonsourcedatacanhelptounderstandthedata,improvedecisionmakingandprovideamoreobjectivepreviewoftheproblemrepresentedbydata(Yau2013).
Agraphiccanalsorevealhiddenpatternsandrelationships.
Visualisationmethodshavegonefarbeyondtraditionaldatapresentationwithsimplechartsandgraphs.
Moderntrendsapproachdatavisualisationasbothascienceandanart.
Ofcourse,certainstandardsofcorrectness(e.
g.
byK.
Mack(*)DepartmentofGeoinformatics,PalackUniversityOlomouc,Olomouc,CzechRepublice-mail:karel.
macku@upol.
cz#TheAuthor(s)2020V.
Pásztoetal.
(eds.
),Spationomy,https://doi.
org/10.
1007/978-3-030-26626-4_8195choosingamethodaccordingtothecharacteristicsofthedata)arestillkept,butthereisanefforttomaketheresultinterestingandcatchytoattractthereader'sattention.
Sophisticateddatavisualisationandinfographicsmethodsofferavarietyofexcitingchartsanddiagrams.
Theadvantageoftechnologiestodayisthepossibilityofpresentingoutputsintheformofonlineinteractivewebtools,whichmakestheprocessingoftheinformation,thatauthorattemptstocommunicate,evenmoreintuitiveandattractive.
Inthischapter,anon-spatialdataanditsvisualisationarediscussed.
Non-spatialdataplaysanundeniableroleintheeldofeconomicsandbusinessintelligence.
Forthatreason,anoverviewofmostcommonandpowerfulpossibilitieshowtovisualiseitwillbepresentedonthefollowingpages.
8.
1SoftwareNowadays,avarietyofsoftwareiseasilyavail-able,knowledgeofsomeofthemisapartofgeneraldigitalliteracy.
Almosteveryone,whosomehowusesacomputer,isabletocreateanyvisualisationusingsomeoftheavailablesoft-ware.
MostofthecomputerusersareskilledwithMicrosoftOfceExcel–softwarethatdoesn'tneedtobepresented(oritsopensourcealternativesLibreOfce/OpenOfce).
Workinginthesetoolsisrelativelyconvenientandstraight-forward,asunderlyingdataandgraphicaltoolsareintegratedintooneuserenvironment,andthewholeprocessisveryintuitive.
However,thisapproachdoesnotalwaysofferproperorhigh-qualitygraphicaloutputsandsupportstheuser'stendencytoblindlyinsertdataintotheprovidedgraphicstemplateswithoutdeeperthinking.
Bythisapproach,datalosesitsabilitytointerpretthestorythatisstoredinit(NussbaumerKnaic2015).
Anotherpointisthetechnicalmaturityoftheoutput.
Intheworldofmoderntechnologies,wheremostoftheinformationisdistributedonline,itismuchmoreprofessionaltoproduceoutputsthatofferadegreeofinteractivityandsupportsimpledistributioninthedigitalenviron-ment.
Interactivityallowstheviewertoengagewithyourdatainwaysimpossiblebystaticgraphs.
Withaninteractiveplot,theviewerscanzoomintotheareastheycareabout,highlightthedatapointsthatarerelevanttothemandhidetheinformationthatisnot(Barter2017).
Forthisreason,sometoolswillbeintroducedofferingthepossibilityofcreatinginterestinggraphicaloutputs.
8.
1.
1TableauSoftwareThecompanyTableauSoftwareoffersasetoftoolsofthesamenamedesignedforexploratoryanalysisanddatavisualisation.
Theproductisespeciallyfocusedonaneffectiveandhighlyaestheticlevelofvisualisation,whichundoubt-edlyattractsmanycustomers.
Thefullversionoftheprogramispaid,buttheversionPublicTab-leauisfreelyavailable.
Inthisversion,ausercanworkwithmanyformatssuchasMicrosoftExcel,Access,textles,JSONles,databasesandalsospatialdata(severaldataformatsaresupported).
Afterloadingthedata,theusercaneasilyselecttheattributestheywanttovisualiseandbasedonthedatatype,asetofoptionsisautomaticallyofferedtocreateavisualisation.
ThemainideaofthePublicTableauistheinteractivityandpresentationoftheoutputsintheonlineenviron-mentsotheresultcanbesharedwithotherusersasattractiveinteractivedatavisualisation.
Thetoolsare,ofcourse,multiplatform,theycanbeusedasadesktop,mobileoronlineversion.
8.
1.
2HTML,JavascriptandCSSHTML,JavascriptandCSSarethebasisofeverywebpage.
WithmoderntechnologiesrepresentedbyHTML5,advanceddatavisualisationrunningnativeintheinternetbrowsercanbedone.
Thissolutionisprobablysuitableonlyfortechnically-advancedusers/developers,whocanhandlecod-ingwiththesetechnologies.
Thereareseverallibrariesdesignedforbuildinginteractive/staticwebvisualisations,forexample,JavascriptlibrariesD3.
js,Charts.
jsorFusionCharts.
Theselibrariesofferdozensofcharts;detailedinforma-tioncanbefoundontheirwebsites.
196K.
Mack8.
1.
3RItisfreeandopen-sourcestatisticalandmathe-maticalcomputingsoftware,primarilyfocusedondataanalysisandmodelling.
SinceRhasbeendevelopedmainlyforstatisticalanalysis,ithasasolidbackgroundfordifferenttypesofcalculationssuitablefordataanalysis.
Thereisalotofpackages,whichcanextendthefunctional-ityofRsoftwarewithjustasimplecodecom-mand.
Thankstothepackages,Risaverymightytoolfordatavisualisation.
Ofcourse,aknowl-edgeofcodewritingisrequired(aswellaswithHTML),whichmakesRformanypeopleinap-plicable.
Butoncethisobstacleisovercome,anewworldofdatahandlingandvisualisationisopened.
Allgraphicscanbesavedinvectorformats,soitispossibletoeditandrenethedesignoftheoutputsinsuitablegraphicalsoft-ware,likeAdobeIllustratororInkscape.
Exceptfortraditionalstaticgraphic,alsointeractiveoutputscanbeproducedwithspecialRpackages.
Sometimes,theinteractivityisredeemedbycom-plexityintheformofoneextralineofcode!
8.
1.
4DatawrapperDatawrapperisanonlinetoolformakingtheinteractivecharts.
Ithasaverysimpleinterface;ausercanuploaddatafromaleorpastethevaluedirectlyintotheeld.
Thetoolgeneratesgraphicsautomatically;ausercanchooseoneofthe16typesofvisualisation.
Severalreningstepscanbedone,likecustomisingofaxis,label-lingorcoloursetting.
Thistoolisanidealsolutionwhenoneneedsaquick,simpleinteractivevisualisationwithoutanyprogramming.
Theseexampleswerejustasmallsliceofwhatnowadaytechnologiesoffers.
Everyoneiscom-fortablewithadifferentlevelofchallenge,con-tentcontrolandoutputoptionssoeveryonecanndtheiroptimaltoolforcreationofgraphicaloutputs.
Thereisanoverviewofanothertoolsforvisualisationinfollowingtable.
Ofcourse,thislistisnotcomplete,therearedozensoftoolsinoffer(Table8.
1).
8.
2ChartsClassificationTheremightbeaconfusioninterminologyregardingthevisualisationofnon-spatialdata.
Usually,words'chart'or'graph'aregenerallyusedtodescribeanyvisualoutput.
Formanypeople,thesetwotermsmeanthesame,butthereisadifference.
Achartisasuperiortermforagroupofmethods,howtopresentinforma-tion.
Agraphisaparticulargraphicaltool,whichshowsamathematicalrelationshipbetweensetsofdata(Blaettler2018).
Withthisapproach,agraphisasubcategoryofachart.
Forthisreason,thetermchartwillberatherusedinthischapter,tokeepthedescriptionofdifferentmethodsmoreboard.
Table8.
1AnexampleofvisualisationtoolsNameOutputDifcultyPricingPlotlyInteractive/staticCodingrequiredFreewithsomepaidplansHighchartsCloudInteractiveEasytohandleFreeD3.
jsInteractiveCodingrequiredFreeCharts.
jsInteractiveCodingrequiredFreeInforgramInteractive/staticEasytohandleFree,extrafeaturespaidRAWGraphsInteractive/staticEasytohandleFreeDataHeroInteractive/staticEasytohandlePaidVisuallyInteractive/staticEasytohandlePaidVismeInteractive/staticEasytohandleFreeGooglechartsInteractiveEasytohandleFreevVizualize.
meStaticEasytohandleFreeSource:Author8Non-spatialVisualisation197Differenttypesofchartswillbedescribedinthefollowingchapter.
Sincetherearedozensofpossibilitiesofvisualisations,onlythemostinter-estingormostcommonlyusedvariantswillbeintroduced.
Forbetterthematiclogic,theindivid-ualmethodsweredividedintothematicgroups.
TheinspirationforthissystemwasthebookVisualizeThis(Yau2011)andthewebsitewww.
datavizproject.
com(FerdioApS2017).
8.
2.
1TrendOvertheTimeTimeseriesaretypicaldataformanyphenomena.
Thingsarechangingintime,andthischangecanbeeasilycapturedandpresentedbysuitablegraphics.
Talkingabouttimeseries,userstrytoexplorethetrendindata.
IsthevalueofthephenomenaincreasingordecreasingArethereanyrepetitivecyclesTemporaldatacanbedividedintodiscreteandcontinuoustypes.
Theknowledgeaboutthischar-acterofdatashouldguidetheuserinadecision,whichkindofgraphshouldbeused.
Forexample,amonthlyrevenuereportisaninformationreferencedtoaone-timestep–amonth,sothiscanbeconsideredasadiscretephenomenon.
Then,asimplebarorpointgraphcanbeused.
Thesecondtypeisthecontinuousdata.
Thisiskindofinformationwhichcanbemeasuredatanytimeofdayduringanydayoftheyear.
Atypicalexamplecouldbeatemperatureoranothermete-orologicalphenomenon;regardingtheeconomicdata,wecanusestockexchangepricesasanexample.
Thestructureofdataissamefordiscreteandcontinuousphenomena,todistinguishthedifference,theproperwayofvisualisationshouldbeused.
Themostprimitivesolutionistoconnectdiscreetlyplotteddatawithanyline.
8.
2.
1.
1BarChartBarchartsarecommonlyused,whichmeanstheuserdoesn'tneedtolearn',howtoreadthegraph.
Thegraphicelementisarectangularbarwhoselengthrepresentsthevalue.
Thetimeaxiscapturestimepoints,whichhavetobeorderedchronologically.
Theneverybarstandsforonediscretetimepoint.
Finally,therearemanyadditionalwayshowtotunethebargraph,e.
g.
barscanbeplacedhorizontallyorverticallyorsomeofthebarscanbehighlightedbyadifferentcolour(e.
g.
timepointswhenthevalueishigherthansetlimitetc.
)(Fig.
8.
1).
8.
2.
1.
2PointChartPointchartworksonsameprincipleasthebarchartdoes,exceptforusedgeometricalelement–itisasimplepointhere.
Thiscansometimesbemoresuitablesincethepointsdonotrepresentsuchgraphiccontentandloadasbars.
Pointchartisalsoknownasascatterplotwhennon-temporaldataisused.
Itiscrucialtoproperlycreateanaxisrepresentingthevalueofthephenomenon,asthereisnootherwaytondoutthevalue.
8.
2.
1.
3LineChartThelinechartisatypeofchartusedforcontinu-ousdata.
Thebasisofthechartisthesameasthebasisofpointchart.
Thecontinuityisaddedbyconnectionofthispointswithlinesegments.
Thenthechartshowshowdatachangesinthetime(particularvalueisstoredinthepoint),andthelinesegmentscreateafeelingofcontinuity.
Italsobetterpointstothetrendbetweentimemarkers(Fig.
8.
2).
Thereisonlyaminordifferencebetweenthelinechartandthesplinechart.
Theydifferonlywiththewayhowthepointsarelinked.
Whilethelinegraphusesstraightlinesegments,thesplinechartplotsattedcurvethrougheachpointfromFig.
8.
1Barchart.
(Source:Author)198K.
Mackthetimeseries.
Thisprovidesamoresmoothandnaturalcourse(Fig.
8.
3).
Anattractivesolutionforadescriptionofchangesbetweentwoorseveraltimepointisaslopechart.
Itcombinestime-approachwithmul-tipleobservedvariables/categories.
Thishelpstoseedifferencesinthedevelopmentofspeciedcategoriesandalsotherateofchangeinoneparticularcategorycomparedtoothers(representedbytheslopeoftheconnectingline).
Atthesametime,deviationsinthegeneraltrendcanbeperfectlyobserved(Fig.
8.
4).
8.
2.
1.
4StepChartLastmodicationofthelinechartisastepchart.
Thisoneisformedbysteppedlinesbetweenthetimepoints.
Itisappropriatetouseitinthesitua-tion,whenthedatarepresentsasuddenchangeinirregulartimeintervals,forexample,apriceofanycommoditywhichhasbeenthesameforalongtime,theninonedaythepriceincreased(Fig.
8.
5).
8.
2.
1.
5GanttChartThischartvisualisesviabarsdurationofseveralcategoriesinatimeseries.
Itillustratesthestartandendpointofoccurrenceofanyactivity/phe-nomenon.
Thischartistypicallyusedasaprojectmanagementtoolforagraphicalrepresentationofthesequencesofactivitiesovertime.
Tasksoractivities,whicharepartsofthewholeproject,aredisplayedinthetimecontext(Fig.
8.
6).
Fig.
8.
2Linechart.
(Source:Author)Fig.
8.
3Splinechart.
(Source:Author)Fig.
8.
4Slopechart.
(Source:Author)Fig.
8.
5Stepchart.
(Source:Author)8Non-spatialVisualisation1998.
2.
2ProportionsProportiondataisgroupedbycategories/types.
Eachcategoryrepresentsapossibility,whichispartofthecertainunit.
Thisdistributionofproportionsisthemostimportantinformationforcomparinggroupsbetweenthemselves.
Withproportionalvisualisation,questionslike"AreallofthecategoriesequallyrepresentedIsthereanycategorywhichdominates"canbeanswered.
Forthistypevisualisation,adataneedstohaveaformofproportionsthataddupto1or100%.
Everypartcouldbestoredrelatively(asapropor-tion)andabsolutely–totalvaluesallowtocom-parenotonlyproportionalpartbutalsototalsize/amountindifferentcategories.
8.
2.
2.
1PieChartApiechartisoneofthemostoftenusedchartsandistypicalforanexplanationofproportions.
Thecirclewhichisrepresentingthewholeisdividedintosectors.
Thearclengthofeachsegment(orinteriorcentralangle,orarea)isillustratingtheproportionofindividualcategories.
Allcategoriestogethermustformaunit/100%.
8.
2.
2.
2DoughnutChart(Fig.
8.
7)Thedoughnutchartisjustamodicationofapiechart,onlytheblankcentreisadded.
Thatallowspresentingofmultipleinformationatthesametimesincetheinnerblankspacecouldbelledwithadditionalrelateddata.
Accordingtosomeoftheresources(NussbaumerKnaic2015),thepieordoughnutchartisaninappropriatewayhowtovisualiseproportionaldata.
Thisiscausedbythegreaterdifcultyofperceivinganglesorareathandistances(whicharethekeyinformationregard-ing,e.
g.
barcharts),itisacommonpropertyofhumaneyeperception.
Inasituationwhentwoormorecategoriesarerepresentedbyanapproxi-matelysamevalue,it'sdifculttodecidewhichoneisgreater.
Thisissuecanbesolvedbyaddinglabels.
Still,severalauthorsrecommendusingdifferentproportionalmethods,likeastacked/simplebarcharts.
8.
2.
2.
3StackedBarChartInsteadofpie/doughnutcharts,simplebarchartorderedfromhighestvaluetoleastcanbeused.
Allbarshavethesamebaseline;theendpointiseasiertocompare.
Evensmalldifferencescanbedistinguished.
Thelengthofbarsisrecalculatedinthatwaythattheirsumequalstowhole/100%.
Astackedbarchartisaperfectsolutionforvisualisingproportionandcomparingseveralclassesatthesametime.
Becauseoftheirgeomet-ricalrepresentation,theyareevenmorespace-savingthanpiecharts.
Stackedbarchartcontainsmultiplevaluesontopofeachother,whichshowsthedivisionofthewholeintocategories.
Concur-rently,individualbarsrepresentthedifferentlevelofcategoriesoreventimepoints.
Forexample–thestackedbarchartcanrepresentsalestrategies:everybarsigniesaparticularstrategy(A-EintheFig.
8.
8),differentcolourshadesrepresentatypeofproduct,andonthey-axis,totalsalesaredisplayed.
8.
2.
2.
4TreeMap/AreaChartThistypeofchartsusesastructureofrectanglesandtheirareatoexpresstheproportionofthewholepart.
Sizeofeveryrectanglerepresentthemetrics.
Theouterrectanglerepresentsparentcategories,andrectangleswithintheparentaresubcategories(Yau2011).
Therefore,primaryrequirementisthatdatahastohaveatree-basedstructure(Fig.
8.
9).
Thereisasimilaralternative,whichdoesn'trequiretreestructureintheinputdataset.
Simplesquareareachart,alsocalledawafechart,usesaregulargridofsmallcells.
Ifthevalueofthecellisset,thentheproportionisexpressedbyanum-berofcells(Fig.
8.
10).
Fig.
8.
6Ganttchart.
(Source:Author)200K.
MackRegardingthetreemaporareacharts,thereisthesameissuewiththeperceptionoftwo-dimensionalobjectaswasdiscussedinpiechartparagraph.
Inthiscase,iftheareamaphasacell-basedregularstructure,theperceptionofinformationcanbedonecorrectlybysimplecountingofcells.
NussbaumerKnaic(2015,p.
59)describesanothersituationwhenareachartsarequitehelpful:"whenvisualisationofnumbersofvastlydifferentmagnitudeisneeded.
Theseconddimensionyougetusingasquareforthis(whichhasbothheightandwidth,comparedtoabarthathasonlyheightorwidth)allowsthistobedoneinamorecompactwaythanpossiblewithasingledimension".
8.
2.
3RelationsandCorrelationTherearemanywayshowtoquantifyrelationsbetweenseveralvariables/group.
Astatisticalapproachprovidesmathematicaltools,suchascorrelationorregression(ifaconditionsregard-ingthecharacteristicsofvariablesarefullled).
Sometimesitismucheasierjusttoplotthedatatorevealthehiddenrelations.
Acorrelationsimplydescribes,howtwovariableschangetogether.
Sometimesitisforgottenthatcorrelationdoesn'tequalcausation.
Basiccorrelationoftwovariablesexpressedwithchartcanquicklydescribethebehaviourofthedata,arateofrela-tioncanbeestimated,maybeaclusteringten-dencycanbediscovered.
Fig.
8.
7Pieanddougnnutchart.
(Source:Author)Fig.
8.
8Stackedbarchart.
(Source:Author)Fig.
8.
9Treemap.
(Source:Author)Fig.
8.
10Areachart.
(Source:Author)8Non-spatialVisualisation2018.
2.
3.
1ScatterplotAscatterplotisoneofthefundamentalchartsusedforplottingofrelationsanddependencies.
ThedataisdisplayedasasetofpointsplacedinaCartesiancoordinatesystem.
Therefore,thechartislimitedfordisplayingofrelationsbetweenonlytwovariables.
Theplacementofpointsinthecharthelpstoeasilyestimatethecorrelationbetweenvariables–iftheyarepositivelycorrelated,pointsareformedintheline-shapedgroup,risingwiththevalueoftherepresentedphenomenon.
Ifthecorrelationisnegative,thislinegrouphasadecreasingtrend.
Withnosignif-icantcorrelation,pointsarenotgroupedintheline-shapedandspreadintheeldrandomly.
Anothercategoricalinformationcanbeaddedintothechartintheformofpointcolourordifferentshapes.
Then,itispossibletoobservedifferencesbetweenparticulartypes;theymighttendtocreateaclusterwhichindicatestheirsimi-larity,orcontroversy,pointsmightbeoverlapping,thenthereisnoclearpatterninthecategoricalgroups(Fig.
8.
11).
8.
2.
3.
2BubblePlotItispossibletoaddathirdvariableintothescatterplotandcomparemoreinformationatthesametime.
Thesizeofthebubblesexpressesthethirdvariable–themeasurehereisanarea,notradiusnordiameter.
Theonlyareacanaccuratelyrepresentdifferencesrelatedtooriginalnumber:ifthedisplayedvalueisdoubledthananother,theareaofagraphicalelementmustalsobedoublesize.
Ifanothermeasureisused(e.
g.
diameter),theratiobetweenvalueandtheareaofagraphicelementwouldn'tbethesame.
Ofcourse,thechartcanbemodiedregardingshapes,squaresortrianglescouldalsobeused.
Attentionmustbepaidtothepositionofaparticulargraphicsele-ment–thebiggerelementmustnotoverlapthesmallerone,soitwouldbenotvisible.
Thisruleshouldbeimplementedinsoftware(Fig.
8.
12).
8.
2.
3.
3ScatterplotMatrixIncasethatexplorationofmorethanthreevariablesisrequired,ascatterplotmatrixisasolution.
Inthiscase,everypossiblecombinationofpairsisplottedbyasinglescatterplot;subse-quentlytheyareallorganisedintoamatrix.
Thisvisualisationcouldbearststepinanexploratorydataanalysiswhentheanalysthasnocluewhatisthedataaboutandwhatisitsbehaviour.
Withanincreasingnumberofvariables,interpretationofthatkindofgraphicpresentationismorecompli-cated,andtheinformationisstillkeptontheelementarylevelofvariablepairs(Fig.
8.
13).
8.
2.
4DifferencesandComparisonComparingasinglevariableisnotademandingtask,thevalueofeveryrecordisdisplayedbyoneoftheprevious-mentionedmethodsandanalysed.
BarchartsorsimplepointchartsmaywellserveFig.
8.
11Scatterplot.
(Source:Author)Fig.
8.
12Bubbleplot.
(Source:Author)202K.
Macktothistask.
Consideringtwoorthreevariables,severalchartsforthistypeofvisualisationhavebeenintroducedintheprevioussections.
Regard-ingthedatawithmorevariables,knownasmulti-dimensionaldata,differentgraphicalmethodshavetobeused.
8.
2.
4.
1HeatmapAheatmapisasimplewayhowtolookatalldataatonce.
Informationisdisplayedinamatrixofregulargraphicalelements,mostlyrectanglesorsquares.
Thevalueforeveryrecord'sattributeisindicatedbycolourintensity.
Thesizeoftheheatmapisdenedbythenumberofrowstimesthenumberofattributes,heatmaphasthesamenum-berofelementsastheinputtabledoes.
Thistypeofvisualisationisnotsophisticatedforthereadingofaccuratevaluesinaparticularrecordbutprovidesagreatoverallviewonthecompletedataset.
Somecharacteristicpatternsindatacanberevealed,e.
g.
tendencytoclustering(Fig.
8.
14).
8.
2.
4.
2ParalelCoordinatesParallelcoordinatesisanothercommonchartforvisualisationofmultidimensionaldata.
Thenum-berofattributesdenesanumberofusedverticalaxes–everysingleofthemrepresentsonevari-able.
ItfollowsthatdifferentaxeshaveadifferentFig.
8.
13Scatterplotmatrix.
(Source:Author)Fig.
8.
14Heatmap.
(Source:Author)8Non-spatialVisualisation203scale.
Toavoidlabellingandaddingmoregraph-icalballastintothechart,datacanbescaled;thenallaxeshavethesamescale.
Parallelcoordinatesaresuitableforrevealingsimilaritiesbetweenrecords.
Forthatreason,labelsarenotalwaysnecessary,itisenoughiftheplotcandenegroupsofrecordswithasimilarpatternortrendontheobservedvariables(Fig.
8.
15).
8.
2.
5StatisticalCharts8.
2.
5.
1HistogramAhistogramisaspecicstatisticalchartwhichdescribesthefrequencyofoccurrenceofvalues.
Thegeometricelementisabaragain,theheightofthebarrepresentafrequency,i.
e.
thenumberofoccurrencesinthecategory,whichbelongstothebar.
Thecategoryheredoesn'tstandforadifferenttype,itratherdenestherangeofvalues,inwhicharedataarebinded.
Itfollowsthatboththehorizontalandverticalaxesarecontinuous(Fig.
8.
16).
8.
2.
5.
2DistributionPlotAlthoughthehorizontalaxisofthehistogramiscontinuous,thedistributionisstilldividedintointervals.
Iftheintervalsizeisnotsetproperly,alotofinformationaboutinter-intervalvariationislost.
Ontheotherhand,plottingofeverysinglerecordwouldmakethechartmessyandconfus-ing.
Acompromisebetweenthisapproachesisadistributionplot,whichisabletocapturethesmallervariationwithinthedistributionandalsosmoothenthedetailedoriginaldata.
Theverticalaxisrepresentstheprobabilityofoccurrenceofvaluefromthesamplepopulation.
Theareaunderthecurvehastobeequaltoone(or100%)(Fig.
8.
17).
8.
2.
5.
3BoxplotAboxplotisanimportantgraphicaltoolfordescriptivestatistics.
Inonepicture,itcandescribeseveralnumericinformation–median,rstandthirdquartiles,andminimumandmaxi-mum(sometimesminimumandmaximumisreplacedbythevaluecalculated:mean+1.
5interquartilerangeoflower/upperquartile).
Outliers(valuesoutofthisrange)areplottedaspoints.
ThespacingsbetweenpartsofboxplotsdescribehowtherawdataisdispersedorFig.
8.
15Parallelcoordinates.
(Source:Author)Fig.
8.
16Histogram.
(Source:Author)204K.
Mackcondensed.
Withboxplots,severalgroupsofdatacanbequicklycompared(Fig.
8.
18).
8.
3AGoodDesignInthebeginning,thedataanalysthastoknowthedataindetail.
Oncetheanalystunderstandswhatkindofinformationishiddeninthedata,whatisthedatatypeandcharacter,hecandecidewhichtypeofchartisthebestsolutionforpropervisualisation.
Thenanotherstepofchartdesign-ingfollows.
Therawdefaultoutputfromthesoftwareisnotwrong,butusually,itisalsonotthemostattractiveresult.
Withanadditionalimprovementofthegraphics,informationwhichtheauthortriestodeliverwiththechartmightbeeasiertoperceive.
Itmustbealwaysconsidered,whoistheaudi-ence,thereaderofthecreatedchart,forwhichpurposeisthechartcreated.
Bydesign,theauthorcanmanipulatewiththewayhowthechartisread.
Ifitwantstofocusonasignicanttrend,theaxisandlabelscanbede-emphasisedwithgreycolour,andtheprimarytrendlineishighlighted.
Then,thetrendistheinformationwhichdrawstheattentionatrst.
Ontheotherhand,achartdesignedwiththepurposeofreadingexactvaluesmusthavereadableandaccuratelabelsofallaxes.
Generally,somerecommendationcanbemade.
Mostlymodestcoloursshouldbeused.
Someofthecolourschemescanevokeemotionsorfeel(e.
g.
redcoloursindicateactivitythatshouldbeaddressed;neutralpastelcoloursmeansthatallfeaturesinthechartareequaletc.
).
Properlabellingshouldbedone–throughusermightknowthecontextinwhichisthechartplaced,hedoesn'tknowthemeaningofeverysingleelementofthechart.
Therefore,atitleofthechart,axesnamesandvaluelabelsandlegendwithexplanationsofcoloursshouldbeapartofthevisualisation.
Geometricalaspectsarealsoimportant.
Sometimesit'smoresuitablejusttorotatethechart,whatmakesitmucheasiertoread(e.
g.
barchartwithlongcategorynames–rota-tiontohorizontalismorenatural,becauseitfollowsthewayhowwereadthecommontext).
Adifferentspatialarrangementofgeometricalfeaturescansolvetheissueswithblankspaceorcantbetterintoawholegraphicdesign(textorposter).
Transformingthegeometricalelementsintopseudo-3Danddisplayingdatathatwayshouldbeavoided(theonlyexceptionisplottingathree-dimensionaldatawitha3Dplot).
Unfor-tunately,forexample,visualisationofthepiechartin3Disquitepopular.
Asdiscussedabove,thepiechartisnotalwaysagoodchoiceforvisualisationofproportionaldata;thecombi-nationwith3Dmakesitmuchmoredifculttoreadorcomparewithotherspiechartsbecausethethirddimensionisproblematicforperception.
Aperspectivein3Dvisualisationcanalsobemisusedforpromotion–asegmentofpiechartplacedintheforegroundlookslargerthanaseg-mentofsimilarsizeinthebackground.
Fig.
8.
17Distributionplot.
(Source:Author)Fig.
8.
18Boxplot.
(Source:Author)8Non-spatialVisualisation205ReferencesBarter,R.
(2017).
InteractivevisualizationinR.
http://www.
rebeccabarter.
com/blog/2017-04-20-interactive/.
Accessed13Nov2018.
BlaettlerK(2018).
Thedifferencebetweencharts&graphs.
Sciencing.
https://sciencing.
com/difference-between-charts-graphs-7385398.
html.
Accessed2Jan2019.
FerdioApS.
(2017).
Datavizproject.
https://datavizproject.
com/Friedman,V.
(2008).
Datavisualizationandinfographics.
SmashingMag.
NussbaumerKnaic,C.
(2015).
Storytellingwithdata.
Hoboken:Wiley.
TableauSoftware.
(2018).
WhatisdatavisualizationAdenition,examples,andresources.
https://www.
tableau.
com/learn/articles/data-visualization.
Accessed30Oct2018.
Tukey,J.
W.
(1977).
Exploratorydataanalysis.
Reading:Addision-WesleyPublishing.
Yau,N.
(2011).
Visualizethis:Theowingdataguidetodesign,visualization,andstatistics.
Indianapolis:Wiley.
Yau,N.
(2013).
Datapoints–Visualizationthatmeanssomething.
Indianapolis:Wiley.
OpenAccessThischapterislicensedunderthetermsoftheCreativeCommonsAttribution4.
0InternationalLicense(http://creativecommons.
org/licenses/by/4.
0/),whichpermitsuse,sharing,adaptation,distributionandreproductioninanymediumorformat,aslongasyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreativeCommonslicenceandindicateifchangesweremade.
Theimagesorotherthirdpartymaterialinthischapterareincludedinthechapter'sCreativeCommonslicence,unlessindicatedotherwiseinacreditlinetothematerial.
Ifmaterialisnotincludedinthechapter'sCreativeCommonslicenceandyourintendeduseisnotpermittedbystatutoryregulationorexceedsthepermitteduse,youwillneedtoobtainpermissiondirectlyfromthecopyrightholder.
206K.
Mack

CloudCone(20美元/年)大硬盘VPS云服务器,KVM虚拟架构,1核心1G内存1Gbps带宽

近日CloudCone商家对旗下的大硬盘VPS云服务器进行了少量库存补货,也是悄悄推送了一批便宜VPS云服务器产品,此前较受欢迎的特价20美元/年、1核心1G内存1Gbps带宽的VPS云服务器也有少量库存,有需要美国便宜大硬盘VPS云服务器的朋友可以关注一下。CloudCone怎么样?CloudCone服务器好不好?CloudCone值不值得购买?CloudCone是一家成立于2017年的美国服务...

野草云提供适合入门建站香港云服务器 年付138元起 3M带宽 2GB内存

野草云服务商在前面的文章中也有多次提到,算是一个国内的小众服务商。促销活动也不是很多,比较专注个人云服务用户业务,之前和站长聊到不少网友选择他们家是用来做网站的。这不看到商家有提供香港云服务器的优惠促销,可选CN2、BGP线路、支持Linux与windows系统,支持故障自动迁移,使用NVMe优化的Ceph集群存储,比较适合建站用户选择使用,最低年付138元 。野草云(原野草主机),公司成立于20...

Hostodo商家提供两年大流量美国VPS主机 可选拉斯维加斯和迈阿密

Hostodo商家算是一个比较小众且运营比较久的服务商,而且还是率先硬盘更换成NVMe阵列的,目前有提供拉斯维加斯和迈阿密两个机房。看到商家这两年的促销套餐方案变化还是比较大的,每个月一般有这么两次的促销方案推送,可见商家也在想着提高一些客户量。毕竟即便再老的服务商,你不走出来让大家知道,迟早会落寞。目前,Hostodo有提供两款大流量的VPS主机促销,机房可选拉斯维加斯和迈阿密两个数据中心,且都...

fusioncharts为你推荐
电子行业动态跟踪报告contentgoogle绑定ipad支持ipad城乡居民社会养老保险人脸识别生存认证深圳市富满电子集团股份有限公司win10445端口Win10系统开放端口号怎样查看?itunes备份itunes 里面的资料如何备份?win7telnetwindows7旗舰版中telnet在哪重庆电信网速测试如何测量网速
深圳虚拟主机 中文国际域名 国外vps 怎么申请域名 美元争夺战 tk域名 申请空间 上海域名 七夕促销 me空间社区 佛山高防服务器 服务器干什么用的 免费cdn 香港亚马逊 linode支付宝 韩国代理ip 秒杀品 畅行云 lamp兄弟连 万网主机 更多