WIXLIB

Beautiful Data in the Eye of theBeholderData Visualization Best PracticesDavid StodderDirector of Research for Business IntelligenceTDWIMay 21, 2014

Sponsor2

SpeakersDavid StodderResearch Director,Business Intelligence,TDWIMichael SaucedaProduct Marketing Manager,IBM3

Agenda The exciting age of beautiful data Goals: why data visualization is important Data visualization and business intelligence– How visualization furthers self-service BI/analytics Who needs what: Research views of visualization use– Reporting, alerting, and visual discovery analysis Dashboard functionality and objectives for visualization Best practices and concluding recommendations4

Visualization: Seeing What Data Hides “Graphics reveal data. “The 20 billion or so neuronsIndeed graphics can beof the brain devoted tomore precise and revealinganalyzing visual informationthan conventional statisticalprovide a pattern-findingcomputations.” – Edwardmechanism that is aTuftefundamental component inmuch of our cognitiveactivity.” – Colin Ware5

Visual Innovation and “Beautiful Data” “For a visual to truly bebeautiful, it must go beyondmerely being a conduit forinformation and offer somenovelty.” “When done beautifully,successful visualizationsare deceptive in theirsimplicity, offering theviewer insight and newunderstanding at a glance.”– J. Steele and N. IliinskyBeautiful Data: The StoriesBehind Elegant DataSolutions, T. Segaran andJ. Hammerbacher, O’ReillyMedia, 2009Beautiful Visualization: Lookingat Data through the Eyes ofExperts, J. Steele and N.Iliinsky, O’Reilly Media, 20106

Data Visualization: A New Language Infographics: Changingrole of data in news reportsand public discourse Data science fame:Tracking Twitter feeds andmore to discover trends,make predictions Visualization: Making bigdata and analytics moreconsumable Experiment and share:see www.manyeyes.com The New York Times; 2014 Peter Sullivan/Best of Show Award (print)winner, Malofiej 22nd International Infographics AwardsData scientist Nate Silver’s prediction vs. actual Electoral College outcome in2012 U.S. Presidential election7

Key Goals of Data Visualization Speed to insight: Help usersavoid slogging through dataand dense tabular reports Actionable: Make it easier toconnect insight to action Clear Context: Enable usersto see how analysis fits intobusiness or performancestrategy Amazement: Excite userswith new ways of seeing data8

Visual Data Interaction: An Imperative Get beyond “puttingpretty pictures onnumbers”: Toward a moreimmersive data experience Drill down from withinvisual objects: Being ableto get deeper into the data,more easily than by writingstandard BI queries Explore relationships:See connections acrossheterogeneous sourcesSanctum, Rogue Pictures, 20119

Encouraging Storytelling, New Formsof Collaboration Using visualization tonarrate and provide contextto the data “story” being told Collaboration benefit: easierto share visualizations,along with annotations andrelated charts to tell thewhole story Coming to the point:Highlighting what isactionable for colleagues10

Visualization and Self-Service BI Trends Business-Driven BI &Analytics: Subject matterexperts want to more controlof data access and analysis– Visualization eases the pathfor nontechnical users Satisfying the variety ofusers and requirements:“uniform” enterprise BIapproaches can fall short– Visualization is key topersonalization of the dataand analytics experience

TDWI Research: Diverse RequirementsView of visualization usage patterns in three key areas:Source: “Data Visualization and Discovery for Better Business Decisions,” TDWI Best Practices Report, Third Quarter 201312

Which Functions Need Which Viz Types?Source: “Data Visualization and Discovery for Better Business Decisions,” TDWI Best Practices Report, Third Quarter 201313

Visualization & Display/Snapshot Reporting Snapshots: Scheduledrather than requested ad hoc;users want to personalizebased on roles– Visualizations must be accurateand consistentSource: IBM Cognos 10 KPIs and scorecards:Orienting users toward goalsand objectives– Can users or developers makethe look more exciting using“fun” visuals? Drill down flexibility: CriticalSource: Lucky Voice on-shift dashboard, from TDWI BestPractices Report.14

Operational Alerting: Avoiding Fatigue Situations that demandimmediate attention: watchout for “alert fatigue”– Using color, size, animation,etc., flexible visualization canhelp users prioritize andrecognize sources– Spotting trends and anomaliesin event data streams Time is of the essence:Real (or near real) time vital Mobile devices: form factora visualization concernCredit: www.catchpoint.com15

Visual Discovery and Analysis Fusion: Analytics, test-andlearn data exploration, andadvanced computationmatched with visualization A visual path: datainteraction through filtering,comparing, and correlatingvisual data relationships Business data laboratory:Enabling exploration of who,what, when, why behindevents and transactions16

Visual Discovery Best Practices Guidance is necessary:Self-service and freedomare important, but mostusers need guidance– A “blank slate” with toomany visual options can beintimidating Metadata matters:Common models,hierarchies, dependencymapping, etc., enableusers relate different datasources and metrics Big data access is oftenimportant: visualizationhelps users cope with datatsunamis– Access to social data, rawdata, use of “late binding”queries to seek insights Performance: Ensuredata management support– In-memory and in-databaseprocessing are attractive forvisual discovery to relievedata movement pressures17

Dashboards: Bringing It All Together Visual, role-based view ofactionable information Nexus of self-service BI andanalytics Performance mgmt: visibilityvia access to data-driven,outcomes-oriented metrics Integration at the glass:internal and external data,metrics, content Many types of dashboards –and often many dashboards18

Old and New Visions of DashboardsInitial Dashboards Tabular reports with fewand only simple charts Limited number, variety ofdata sources Limited methods offinding, interacting w/data Dependent on ITdevelopers to create andmodify Tied to single tool orapplicationWhere They Are Going Libraries of chart types;drag-and-drop selection Role-based, single view ofdata from multiple sources Integration of search,advanced analytics tools Self-service creation,preferably managed orguided by IT expertise Integrated view; seamlessexperience on mobile19

Dashboard, Visual Analysis Functionality Filtering and using data tofilter other views highest incurrent use (57%) 45% color-coding data andlinkages to charts Highest “plan to implement”:Data comparison acrossmultiple visualizations (41%),alerts (40%), and drag-anddrop elements (38%) Highest “no plans”: call-toaction buttons (42%)Source: “Data Visualization and Discovery for Better Business Decisions,” TDWI Best Practices Report, Third Quarter 201320

Time Series Analysis: Demand Driverfor Visualization Users need to analyze datachanges over time 39% currently implementingvisualization for time series Getting beyond basic linecharts to bring in more dataviews: scatterplots, 3D Developing complementaryvisualizations for historicaltime series analysis, realtime views, and predictiveanalytics21

Justifying Projects: Benefits SoughtUsing visualization to reduce time to insight has benefitsfor all types of users in many different scenariosSource: “Data Visualization and Discovery for Better Business Decisions,” TDWI Best Practices Report, Third Quarter 201322

To Start: Begin with the Data “All of this comes togetherto paint you a picture of astory that is in fact alreadythere in the data – but ifyou don’t have the rightlens to see it, you can’tsee it.” – Matt Felton,Datastory ConsultingGeospatial analysis of potential for Planet Fitness clubmembership cannibalization. From TDWI Best Practices Report “Most visualization storiesbegin with some kind ofquestion that orients theviewer to the topic andcontext within which thedata is most meaningful.”– Steele and Iliinsky What data are we lookingat? In what time frame doesthe data exist? What notable events orvariables influenced thedata?23

Choosing Visualizations: Best Practices Avoid clutter; no “eyecandy” Consider the audience:executive? A team? Pay attention tocontext; emphasizewhat matters Aim for relevance; don’tmislead or confuse Step beyondconvention – but do sowith purposeSource: “Data Visualization and Discovery for Better Business Decisions,” TDWI Best PracticesReport, Third Quarter 201324

Concluding Recommendations Improve data visualization and visual analysis fornontechnical users– “Nontechnical” users struggle to interact effectively with data– Visualization can give them easier and more powerful interaction Match visualization capabilities to users’ types ofactivities– Display, snapshot reporting, or scorecards?– Operational alerting?– Visual data discovery and analysis? Consolidate interfaces and use dashboards for singleview– Dashboards can provide a complete and consolidated interface– As self-service BI/analytics expands, data views can proliferate;organizations should seek to reduce chaos and complexity25

Concluding Recommendations Increase data interactivity with broader visualizationfunctionality– But rather than give users a blank slate, ensure that they have guidance,either through the software or from IT developer assistance Ensure you have data management and performancestrategy for visual discovery and analytics– In-memory and in-database computing can be a valuable components ofa data architecture to support complex, compute-intensive analytics withless data movement Create “beautiful” data visualizations by reducing clutterand increasing speed to insight– Experiment with visualization libraries to discovery what willbest express insights and make them actionable26

Thank You!David StodderDirector of Research for Business IntelligenceTDWI (www.tdwi.org)dstodder@tdwi.org(415) 859-993327

Imagine if you could delight endusers GUIDED creation withrecommendation on bestvisualization for your dataENGAGING with a focus on enduser interactivity and LOBconsumption appealEVERYWHERE with ability tointeract with your visualizationanytime anywhereFLEXIBILE with ability toincorporate new and compellingvisualizations as needed

IBM SolutionRAPIDLY ADAPTIVE VISUALIZATION ENGINE A simple descriptive language to describea chart Flexible enough to describe all knowncharts and extensible to describe newand innovative visualizations

IBM SolutionRAPIDLY ADAPTIVE VISUALIZATION ENGINE

RAVETHE ENGINE DRIVING IBM VISUALIZATIONSAnalytical DecisionManagementSPSS AnalyticCatalystCognos Business IntelligenceCognos Insight

IBM Watson AnalyticsNEXT GENERATION OF DATA DISCOVERY

IBM CognosVISUALIZATION CUSTOMIZER

RAVECREATING NEW VISUALIZATIONS FOR THEMARKETPLACEThe old way The new way Analytics & RAVE Visualization EngineAnalytics & Visualization yVisualizationDescription

Enabled by an unmatched analyticsportfolio IBM Business sBusinessImperativeBusinessIntelligence &Predictive ORATESimulateScoreGovernSurveyDecideUtilize and combine data toderive a complete pictureTransform systems ofengagementDeliver trusted risk intelligenceacross the organizationOperationalize analytics todrive decisionsSpan finance, sales andoperationsOptimize risk and reward for thebusinessEnable forward lookingBusiness IntelligenceInfuse scenarios andpredictive intelligenceSupport smarter, risk-awaredecisions

The IBM DifferenceCreate rapid, moreinnovative visualizationsWatson AnalyticsIBM Center for AdvancedVisualization

Questions?37