JUMPING INTO BIG DATA - Researchexcellence
JUMPING INTO BIG DATA:How the Media and Entertainment Industries areGettingStartedOctober 23, 2014#CREbigdata
OPENING REMARKSRichard Zackon, CRECeril Shagrin, CRE Council ChairStacey Schulman, Big Data Chair
ABOUT THE CREThe Council for Research Excellence is a body of senior research professionals,formed in 2005 to identify important questions about audience measurementmethodology and to find, through quality research, the answers to those questions.The Council provides the Nielsen client community a means to undertakeresearch projects no one company could undertake on its own.3
CRE MEMBER COMPANIES4
BIG DATA COMMITTEE MEMBERSStacey Schulman, Chair Michele Buslik Annette Malave Jon Cogan Michele Meyer Laura Cowan Dan Murphy Pete Doe Rosemary Scott Janice Finkel-Greene Ceril Shagrin Sam Garfield Howard Shimmel Paul Hockenbury5
BIG DATA COMMITTEE - MISSION STATEMENT The challenges of traditional, sample-based market researchcontinue to rise in our dynamic, fragmented media landscape. At thesame time, accessibility to large data sets from almost any businessor personal sector (including our own physiological responses) isdriving increased demand and innovation within the data sciences. The goal of the Big Data Committee is to explore the growingintersection of these two disciplines, identify and create informeddialogue around the critical questions this intersection creates, andexplore methods, techniques and approaches to improve the qualityof big data solutions.6
VENDOR PERSPECTIVE: Under The Hood OfBig DataGeorge Ivie, CEO, Executive Director, MRCKen Barbieri, VP Market Development, NeustarAndrew Fiegenson, Managing Director, NielsenMainak Mazumdar, Chief Science Officer, SimulmediaNishat Mehta, EVP, Global Partnerships, DunnHumby
BIG DATA CASE STUDY: Integrating CreditCard Transactions And Audience Data ToBetter Understand And Reach ConsumersPete Doe, SVP Data Integration, Nielsen
INTEGRATING CREDIT CARDAND AUDIENCE DATA FORPRECISION MARKETINGPETE DOE
CONTENT1.Background2.The Opportunity3.The challenges – Privacy, Integration Methods4.Validity5.Conclusion
BIG DATA HAS SPARKED AN EXPLOSION OF NEWTECHNOLOGY PROVIDERS
ONLINE AD LANDSCAPEOnline Advertising is increasingly moving toProgrammatic Buying
OPPORTUNITYIncorporating offline activity improves theefficiency of buying and selling
NIELSEN AND PROGRAMMATIC BUYINGNielsen (and partner) dataprovides offline consumer andmedia activityCredit Card Transaction DataTV viewingCPG segments(Nielsen Catalina)Prizm SegmentsCredit Card ActivityLinking these data with onlinedatabases enables moreeffective online advertisingDetailed retailer level purchasesAirlineHotelsApparelMass MerchandiserBaby StoresHome ImprovementBook StoresPet StoresCasual DiningSupermarketsDepartment StoresTravel ServicesFast FoodWirelessFine DiningElectronics
INTEGRATING RETAIL ACTIVITY INTO ONLINE ADSERVINGCredit CardTransaction Data(80% of USTransactions)Nielsen OnlinePanel 200,000CookiesDMP(300M cookies)MultipleProcessesChallenges:a) Privacy ab) StatisticalAd Served toAdtoPC,ServedLaptop,PC,Laptop,TabletorTabletor PhonePhone
ENSURING PRIVACY – TWO PROCESSESCredit CardTransaction DataNielsenOnline PanelSecureMatchingThird PartySecure Match(Names andAddresses)LookalikeModelingIndividuals withcookies and retaileractivity from creditcard dataLookalikeModeling inSecure AnalyticEnvironmentIndividuals withcookies and retaileractivity from creditcard dataIndividuals withcookies andmodeled retaileractivity from creditcard data
LOOKALIKE MODELINGSecure Analytic Environment to Ensure CC data PrivacyIndividuals with cookiesand retailer activityfrom credit card data online behaviorRespondent levelData FusionMatching based ondemographics, onlinebehavior, retail purchasesIndividuals with cookiesand modeled retaileractivity from credit carddata
DELIVERING ADSIndividuals with cookiesand modeled retaileractivity from credit carddataCookie Matching andmodelingModeling uses onlineactivity tracked throughcookieshpDMP(300M cookiestyped with retailsegments)AdNetworks/Publishers
STATISTICAL VALIDITY TESTDMP(300M cookiestyped with als withcookies and retaileractivity from creditcard dataTruthValidation Database
STATISTICAL VALIDITYOur Objective: Demonstrate Precision viaImprovement on Random Measure 125Precision: IndexIndexon Random Ad Serving2982501250135129Total 42%(13%)15415919%11%1981876%Penetration Group2%0.4%
CONCLUSIONMaking Ads relevant is good for the advertising industryand good for consumersIntegrating Data Sources delivers more effective andrelevant online advertisingGood Data, Coverage, Privacy Compliance andStatistical Validity are Essential Elements
THANK YOU!Pete.Doe@Nielsen.Com
BREAK
MARKETER PERSPECTIVE: One Client’s DataJourneyMark Kaline, Former Global Media Director at Kimberly-Clarkand Former CRE Chair
One Client’s Data JourneyCRE Big Data Case Study Event – October 23, 2014Mark Kaline
“By 2017, the CMO will spendmore on IT than the CIO”.Gartner, February 2012
BIG DATA: WHAT IS IT? Big Data is commonly described by the technologyindustry using the 4 V’s: Volume: Refers to the massive amount of data being collectedby companies, through internal and external means Velocity: Refers to the frequency of data generation orfrequency of data delivery (real time) Variety: Refers to the types of data being collected- structured(numbers, URLs, ) or unstructured (video, images, text/chat) Value: Refers to the ability for the data to drive insights whichcould impact effectiveness, efficiency, profitability and growth.
BIG DATA: DEVELOPING A DATA STRATEGY IS AJOURNEY Collecting Data is not enough: Most companies already have a flood of data – “InfObesity” Screen hopping and multi-tasking generating even more data What is it you are hoping to learn? What’s the what? What are other parts of the organization hoping to learn? How does the organization collectively/exponentially learn? Do you have the right data? Is it good? And where is it? What is the plan to fill gaps in data to complete the desiredanalysis? Can it be repeated regularly & automated to speed learning? Does the data drive insights that are actionable? Big Data is Nothing without Big Insights
DEVELOPING AN APPROACH TO INTEGRATEDMARKETING INTELLIGENCE
MOBILESOCIAL3rd Party ID&V DataLBSWEBCALLCENTREMEDIA SDEVELOPING THE TECH CAPABILITY FOR INTEGRATEDMARKETING eCOMM.APPLICATION LAYERTAGGINGBUSINESS/ ENTCONTENT3rd Party CookieData SyncSOCIAL LISTENINGCONTENTCONTENT USINESSBUSINESSRULESBUSINESS RULESBUSINESS RULESBUSINESS RULESBUSINESS GMENTMETADATAMETADATADECISION / NBA ENGINEREPORTINGPLATFORMCUSTOMERSEGMENTATION /MODELLINGCAMPAIGN KPI ALERTSMARKETING DATA WAREHOUSE (SCV)Master DataCustomerDataPIICampaign /Offer DataCompetitive/ MarketIntelligenceProductDataSales &DistributionData3rd Party Data
1st Party Data-The OvenSocial DataSearch DataPerformance DataResearch CRDisplay/SocialVideo3rd Party Data SourcesPrivateMarketplacePre- Bid Safety, Verification, ViewabilityThe OvenMobile3KC Trading Desk Technical Diagram32
CURRENT KCTD DMP STATUSOver 5 million unique 1st party usersIntegrated with outgoing Oven CommunicationsConnected to Social Media ActivityCapturing Brand.com actions and personasContains all Survey respondent informationLinked to all Paid Search activityActionable across all KC programs3
KCTD DMP 2014 ROADMAPQ1Global RolloutDeeper Integration with CRMTest and Learn with Dynamic CreativeQ2S2S Sync with OCR Video ProviderQ3Connecting Offline & OnlineQ4Creation of Always-On Unified Targeting Profile
BIG DATA : INCREASING MEDIA ROI Marketing organizations desire data & attribution tobetter drive decision making Clients can turn big data into a powerful media asset,driving optimization, increased ROI and targetingeffectiveness. Moving from commercial programs that run, tocommercial programs that learn Applying learning, where possible, in real time. Implementing programmatic buying where possible tolearn dynamically. Recycling buy information thru CRM system to drivegreater and greater precision.
CRM TRADING DESK ALIGNMENT Used criteria from CRM database, email, and websites to create unique targeting strategies Developed Strategic targeting and optimization architecture and rules Generated in highest conversion rate of any tactic within Potty Breaks program
DRIVE PRE-SEASON PURCHASE Challenge: Swim diapers are not inseason across the nation at thesame time, so can waste beeliminated by focusing on marketsthat met certain criteria Approach: Via Trading Desk, use NationalWeather Service data feed totarget geo locations that are 70 degrees and sunny Results: Cost-per-action was 13%below goal and survey resultsindicated 67% purchase intent
AVOID THE ONE DAY SCRAMBLE Challenge: Change the mindsetfrom buying a single box of facialtissue during acute symptomologyto buying multiples, prior to theacute need, in an effort to beprepared for the season. Approach: Leverage Google’sCold/Flu data to geo-target ourmedia in areas that have thehighest Cold/Flu symptomologies Via Trading Desk: Copy servedbased on level of flu symptoms inarea Via Social: Re-target those whomention cold/flu in posts
SEM RETARGETINGRetarget consumers searching specific keywords in GoogleThis tactic consistently performs among the best across various KC brandsSEM Retargeting ( 7.52 CPA) outperformed all but 2 tactics in Kotex efforttamponstamponsUser searches related keywordKCTD captures user dataUser targeted with UBK ad39
DATALOGIX TV TARGETINGKCTD utilized data provider, Datalogix, to target consumers with a high propensity to have seen Snug &Dry TV adsProcessConsumer sees Snug & Dry AdSet Top Box Data collected; aligned withonline addressable dataConsumer served Snug & Dry ad onlineResultTV Targeted audience was 59%more likely to purchase DiapersQ. How likely are you to buy Huggies Diapers in the next 30 days?40
BIG DATA CASE STUDY: Can BehavioralData & Machine Learning Algorithms HelpBrands Grow Audience Interactions?Yoram Greener, Founder, JubaPlus
http://bit.ly/P1OgH3CAN BEHAVIORAL DATA AND MACHINELEARNING ALGORITHMS HELP BRANDSGROW AUDIENCE INTERACTIONS?October 23th 2014JubaPlus LLC. - Proprietary
AGENDA The Business Need: capture consumers moments of interests Industry Challenges Case StudyJubaPlus LLC. - Proprietary43
JUBAPLUS – AN OPTIMIZATION AGENCYJubaPlus LLC. - Proprietary44
FULL CYCLE OPTIMIZATIONFROM SPEND TO CONVERSIONSome 57% of marketers agree that data drives higher conversionrates, and 34% said it provides insights into customer behavior -- butmost don't understand how to aggregate numbers from siloed mediasources to drive overall better resultsFailure To Master Online Data Costs Marketers Profitsby Laurie Sullivan, MediaPost Jan 21, s-prof.html?c 103844#reply#ixzz2JDwsBi4y
The Business NeedCapture consumers moments of interestsConsumers consume content across devicesVODDVRDesktopsNetflixHuluAmazonLaptops
The Challenges: Information AccessConsumers search and makedecisions-Faster than marketers can answer-Anywhere 24/7-Based on small butrelevant particles of information
Industry Challenges: Classic EconomicProblem of Demand and SupplyToo much consumer data, too fastConsumer Demand for RelevantInformation @ SpeedBrand Capacity andCostMarketers overloaded, over worked, while supply of contentincreases
Industry Challenges: Classic EconomicProblem of Demand and SupplyFinding the equilibrium point between consumer informationdemand and relevant content supply @ speedMarketers have done sub-effective and deficient jobs: matchingbetween what consumers search and their content servings
CASE STUDY: SOCIAL MEDIA CONTRIBUTIONNovember 2012
Main FindingsCONVERSATIONS AMONG CONSUMERSRESULTED IN XXX OF 3,088 ADDITIONAL XXX (3%OF 88,249 UNITS) AND 14,529 XXX (3% OF 415,130UNITS) OR XXX OF 66.1 MILLION AND 435.8MILLION OVER THE COURSE OF ONE YEAR.Proof of Concept 51
Impact of Earned Media – Conversations among ConsumersIMPACT OF EARNED MEDIA – CONVERSATIONSAMONG CONSUMERSObjective: Quantify the impact of conversations amongconsumers on actual xxxAlternative Media TypesMethodology: Measure major media investments (paid,owned, and earned) between Oct 2009 andJune 2011 on a weekly basis Use marketing mix modeling to isolate theimpact of each media alternative (paid,owned, earned) on weekly sales Evaluate impact of alternative media types: Digital Paid (online including search orSEM and display advertising on automotive sites such as Edmunds or Kelley Blue Book) Traditional Paid (offline advertising, particularly nation-wide network TV) Owned (branded Facebook pages, YouTube videos) Earned (positive and neutral social conversations on social networks and blogs). Evaluate impact of sales channels (xxx events and xx activities, website visits) Earned Media and Owned Media are significant drivers of Social CurrencyProof of Concept 52
Impact of Web TrafficOF TOTAL XXX OF 88,249 XXX AND 415,130 XXX, EIGHTPERCENT WAS ACCOUNTED FOR BY TRAFFIC ON THEWEB, WHILE THE REST WAS DUE TO OTHER EFFORTSSUCH AS XXX EVENTS AND XX ACTIVITIESAverage xxx traffic92%All Other/ BaseProof of Concept 53
Impact of Web trafficON THE AVERAGE, TRADITIONAL PAID MEDIACONTRIBUTED MOST TO WEB TRAFFIC THAT GENERATEDEIGHT PERCENT OF TOTAL SALES; EARNED MEDIACONTRIBUTED 7.3 PERCENT OF WEBSITE TRAFFICAverage xx Contribution100%8%80%60%40%20%0%92%Average Conversation ContributionWebtrafficAll tional PaidDigitalDigital PaidAllOther/BaseOther/BaseBase is the website visits that would be generated without Digital Paid, Traditional Paid, Owned, and Earned (Conversations) Media effortsProof of Concept 54
Impact of Web trafficOVER THE COURSE OF 14 MONTHS, THE CONTRIBUTION OF DIGITAL PAIDMEDIA WAS FAIRLY CONSTANT, WHILE TRADITIONAL PAID MEDIA (SUCH ASNETWORK TV ADVERTISING) DECREASED. OWNED MEDIA (VISITS TOFACEBOOK BRANDED PAGES OR YOUTUBE VIEWS) INCREASED GRADUALLYSINCE NOV 2010. CONVERSATIONS AMONG PEOPLE ALSO INCREASEDGRADUALLY.What DrivesWebsiteTraffic?Total WebsiteVisits onalOfflinePaidOwned (Facebook,Facebook/Youtube/OtherYouTube, other)EarnedConversation*350Website Visits (K)300250200150100500Base is the website visits that would be generated without Digital Paid, Traditional Paid, Owned, and Earned (Conversations) Media effortsProof of Concept 55
Impact of Earned Media – Conversations among ConsumersFINDINGS AND MAIN CONCLUSIONSOverall Impact Three percent of total xxx was driven by earned media (social conversations amongconsumers). With an average car model price of 25,000 for the xxx and 35,000 for thexxx, this means that social conversations contributed about half a billion in additionalsalesRelative Importance of Media Types: Over the time period of this study, Owned Media and Earned Media had increasinglyhigher contributions to web traffic There was a significant increase in Owned Media (web traffic generated from Facebook /YouTube branded pages) that drove social conversations over digital paid and traditionalpaid network advertisingOur main conclusion is that the profits realized from 66.1 million and 435.5 million ofadditional xxx generated from social media channels (owned media on Facebook orYouTube, etc.) or through social conversations significantly outweigh the investments.As far as marketing communications was concerned, the investment to profit ratio of socialmedia was among the most efficient of all media alternativesProof of Concept 56
THANKSYoram Greeneryg@jubaplus.comJubaplus.com@jubaplus
MEDIA PERSPECTIVE: Putting Big Data InThe Media KitShaun Doyle, CEO, Cognitive BoxHoward Shimmel, Chief Research Officer, Turner Broadcasting
CRE JUMPS INTO BIG DATARichard Zackon, Facilitator, CRE
CRE PROJECT REPORTA PREDICTIVE MODEL OF LOCAL TV RATINGSUSING SUPERVISED MACHINE LEARNING.CRE Big Data EventOctober 23, 201460
PROBLEM Over 150 local TV markets are currently measured by Nielsen usingonly a paper diary. CRE has demonstrated that relatively small sample sizes renderaudience ratings unstable and nonresponse bias furthercompromises the accuracy of diary estimates. Broadcasters and advertisers in small markets lack reliable and validmetrics with which to plan and conduct business.61
OPPORTUNITY The Nielsen National People Meter has some 20,000 HHs and is addingan additional 2300 households in diary-only markets. By applying techniques of machine learning, it may be possible to usethese data to estimate local market ratings with accuracy significantlygreater than current diary sample methods.62
CRE LOCAL MEASUREMENT COMMITTEE Under the leadership of Billy McDowell (Raycom Media) the committeeis exploring alternatives to paper diary measurement. They have commissioned a team to explore whether machine learningtechniques can improve the accuracy of ratings estimates.63
RESEARCH TEAM: Vasant Dhar: NYU Stern School of Business. Data Scientist Tim Dolson: LORE Media Research (Formerly VP, StatisticalMethods, Nielsen). Data Consultant Sandy Retsky: Independent. Database Programmer Richard Zackon: Audience Patterns LLC, Project Manager64
SUPERVISED MACHINE LEARNING An artificial intelligence technique in which the computer is presentedwith example inputs and their desired outputs in order to learn ageneral rule, an algorithm that maps inputs to outputs Inputs: TV Household characteristics, TV station characteristics anddemographic viewing in 20,000 People Meter Households fromoutside a local market. Outputs: ratings estimates for the specific local market. Machine learning will develop a fitted model based on a “training set”and assess predictive accuracy with a “test set” of 60 stations.65
RESEARCH PLAN A test of 60 stations in ten Local People Meter markets Demos: HH, P2-17, M18-49, M50 ,W18-49, W50 To predict local ratings by QH ratings for 16 weeks in 2013 usingsimultaneous People Meter data from outside each market Compare predictions with actual People Meter estimates from withinthe market to assess the accuracy of the predictions66
DATA Nielsen has provided tuning and viewing data by QH forHouseholds and persons for all People meter HH’s forFebruary/May/July/Nov 2013. Nielsen has also provided Universe Estimates by market anddemographic, geographic and psychographic data (e.g. Claritas)for each sample household.67
ANALYSIS The research team will prepare the data for analysis by assessingits quality, analyzing distributions and performing appropriatetransformations. This will be followed by assessing various machine learningmethods for the problem and applying the resultant algorithm ondata from respondents with known features. The test metric will be Live Demographic ratings by QH.68
DELIVERABLE The CRE will be provided with a final report which will describe theanalytic process including algorithms, validation statistics andrecommendations for further R&D improvements to be taken up byCRE or Nielsen.69
TIMETABLE September 4: CRE Approval October 6: Final specs October 10: Nielsen provided initial data November 14: Initial progress report to CRE Local Committee December 15: Final report to full CRE70
BENEFITS If the results are successful: Local TV markets can consider an affordable new currency, more stable than onebased on small samples and more valid than one based on diaries or STB’s. Nielsen will be encouraged to consider further refinements to the model wedevelop. The industry will be encouraged to innovate with careful use of advancedanalytics. The industry will have had a front row seat to learn from the project. If the results are unsuccessful or are inconclusive: The industry will have established a benchmark level of predictive accuracy formachine learning techniques, setting the basis for further improvement. The industry will have had a front row seat to learn from the project.71
CURRENT QUESTION:What is the standard of accuracy required?72
INTRODUCING: CRE Big Data PrimerGerard Broussard, Principal, Pre-Meditated Media
BIG DATA: KEY ASPECTS Why Big Data? Structuring the Streams of Big Data Big Data Defined Traditional Market Research Data Quality Data Science Talent Privacy Getting Started with Big Data Marketplace Feedback74
WHY BIG DATA?BIG DATA INTEREST OVER TIMESearch Index for term “Big ce: Google Trends – normalize search volume according to the ratio of term(s) to the entire volume of search
WHY BIG DATA?FUTURE GROWTH OF BIG DATA VOLUME
STRUCTURING BIG DATAData require shaping to enable comparison with structured data likeTV ratings and retail salesSocial media conversationsGeo-location coordinatesMobile app usageSocial media picturesRetail traffic patternsVideo water markingAudio water markingSocial media graphics77
BIG DATA DEFINED: IT’S NOT JUST ABOUTSIZEBig Data in Marketing and Advertising (BDMA): Too big to handle on single file server Most likely includes unstructured data Multiple data sources, reflecting consumer touch points Complexity depends on marketing/advertising question78
BIG DATA & DECISION MAKING Surfaces insights and facilitates feedback immediacy notpossible with traditional analytic/research approaches79
TARGETING INSIGHTS - MARKETER A, BRAND BPOOLING DATA TOGETHER TO IDENTIFY HIGH-PROPENSITYCONSUMERS, AKA THE HEAVY HITTERS FacebookTwitterPinterestOtherCustomerSatisfaction Email/webTelephoneMailCustomerPurchase Volume Price pointFrequencyBasket sizeStore-level Sales Big BoxDept. storeSpecialtyGeographic Skew RegionMarketCounty/ZipSocial MediaUnstructuredStructuredBrand B EstimatedSales Volume als10%80
TARGETING INSIGHTS - MEDIA AGENCYIDENTIFYING EFFECTIVE MEDIA THAT DELIVER HEAVY HITTERS . . .Heavy Hitter IndexChannel IndexUnstructuredSocial MediaTV Data Source FacebookTwitterPinterestOther NetworksDaypartsPrograms3rd Party MatchVehicle IndexPhone App A 108Television115Phone App B 106Internet110Phone App C 99Mobile95Tablet App G 97StructuredInternet SitesSearchRadio87Tablet App E 94Mobile Data TabletPhonePrint82Tablet App F 92Radio FormatMarketPrint Circulation MarketTablet App G 90Tablet App H 8981
FEEDBACK IMMEDIACYALWAYS ON, ELECTRONIC ACCESS OPPORTUNITYFOR ADJUSTING COURSEDaily Sales PerformanceGoalActual1063. Actual Sales ResumeSynch With Goal104102SalesPerformanceIndex1009896941. Actual SalesRunning Behind Goal123456782. Course-CorrectiveAction Begins9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24Day of Mktg Program82
DATA QUALITY GOAL – CLEAN, FIT AND VALIDData integrations or “mash-ups” beg questions of quality andcomparabilityGolf Club Bag ProspectsGolf Club Bag ProspectsCustomer DatabaseExternal DatabasesTV STBAnnual Golf Occasions 12 Annual Golf Travel 2 Medium/Hvy Golf TV ViewerOwn specialty Golf ClubsPacific/Northeast ResidentAge 55 Medium/Hvy SkierOwn High-End SUVDigitalFind The Look AlikesTransactional83
BIG DATA QUALITY: WHAT TO LOOK FOR“Careful inspection of the underlying representativeness, ensuringconsistency or reported metrics over time and understanding how datacollection might impact accuracy.” George Ivie, Executive Director, MediaRating Council (MRC)1. Underlying Data Values2. Time Period3. Representation4. Consistency84
DATA SCIENTISTS: RARE BIRDSThey’re not your traditional media or marketing researchanalyst Ability to organize/work with large data sets Advanced statistical background Recruited directly from academia or outside ad industry In short supply85
GETTING STARTED IN BIG DATAOne part strategy, one part technology Tangible goal and strategy statements TV network – “reduce social media post storage costs by 25%” Marketer – “Uncover new target segments within customer data base andthe touch points to reach them” Technology implementation (examples) Hadoop – enables multi-server processing of large data sets MapReduce – algorithmic framework within Hadoop; “air traffic controller”86
TECHNOLOGIES FOR BIG DATATechnologyDefinitionHadoopOpen-source software for processing big data across multiple parallel servers.MapReduceThe architectural framework on which Hadoop is basedScripting languagesProgramming languages that work well with big data (e.g., Python, Pig, Hive)Machine learningSoftware for rapidly finding the model that best fits a data setVisual analyticsDisplay of analytical results in visual or graphic formatsNatural language processing (NLP)Software for analyzing text—frequencies, meanings, etc.In-memory analyticsProcessing big data in computer memory for greater speedSource: Thomas Davenport, “big data @work,” Harvard Business Press, 201487
PRIVACYPUBLIC CONCERN SPANS VIRTUALLY EVERYASPECT OF BIG DATA% Very Concerned81%80%75%64%Legal Standards TransparencyDataCollection of& OversightAbout Data Use Storage/Security Telecom Data59%58%Collection ofVideo/AudioDataCollection ofLocation DataSource: White House, Office of the President, May 2014, 24,092 respondents88
MARKETPLACE FEEDBACK* Successful Deployment Requirements strategy management support data talent internal education Hottest Areas in Marketing and Advertising targeting and addressability creating consumer multi-touch-point profiles speedy decision making Biggest Implementation Challenges Attracting data science talent integrating data from disparate sources* Ten companies comprised of marketer, media agency, media firm, research/data firm89
MARKETPLACE FEEDBACK – WE’RE AT THEEARLY STAGES“Difficult for unacquainted to understand what they can get from the data. What is thequestion? The push must come from the top.” VP Sales/Analytics, MVPD“Only the most advanced companies have a truly structured plan, detailed byobjectives and data sources. Companies still in early adopter stage, still trying tofigure out what it (Big Data) means.” VP of Partnership Development, Big DataSyndicator“Critical that it’s woven into your business processes so the organization knowswhat to do with it.” COO, Digital DSP“Biggest challenge is finding quality human resources, gathering and reportingintegrated touch point data.” SVP, Director of Analytics, Media Agency90
CLOSING REMARKS
CPG segments (Nielsen Catalina) Prizm Segments. Credit Card Activity. Linking these data with online databases enables more effective online advertising. Airline Hotels Apparel Mass Merchandiser Baby Stores Home Improvement Book Stores Pet Stores Casual Dining Supermarkets Department Stores T
The Rise of Big Data Options 25 Beyond Hadoop 27 With Choice Come Decisions 28 ftoc 23 October 2012; 12:36:54 v. . Gauging Success 35 Chapter 5 Big Data Sources.37 Hunting for Data 38 Setting the Goal 39 Big Data Sources Growing 40 Diving Deeper into Big Data Sources 42 A Wealth of Public Information 43 Getting Started with Big Data .
10 to promote the breeding of high quality Hanoverian show jumping horses in North 11 America. To be selected for the program, mares and stallions must have both an 12 appropriate jumping pedigree and demonstrated jumping ability or offspring with 13 demonstrated jumping ability. 14 Stallion requirements:
Lesson Three: Students fold origami jumping frogs and determine their frog’s jumping rate. Students create tables, graphs, and equations based on their frog's jumping rate. Each students should be given a 3'' x 5'' card and instructions for folding an origami jumping frog. Students should be allowed to help each other with the folding.
big data systems raise great challenges in big data bench-marking. Considering the broad use of big data systems, for the sake of fairness, big data benchmarks must include diversity of data and workloads, which is the prerequisite for evaluating big data systems and architecture. Most of the state-of-the-art big data benchmarking efforts target e-
of big data and we discuss various aspect of big data. We define big data and discuss the parameters along which big data is defined. This includes the three v’s of big data which are velocity, volume and variety. Keywords— Big data, pet byte, Exabyte
Retail. Big data use cases 4-8. Healthcare . Big data use cases 9-12. Oil and gas. Big data use cases 13-15. Telecommunications . Big data use cases 16-18. Financial services. Big data use cases 19-22. 3 Top Big Data Analytics use cases. Manufacturing Manufacturing. The digital revolution has transformed the manufacturing industry. Manufacturers
Big Data in Retail 80% of retailers are aware of Big Data concept 47% understand impact of Big Data to their business 30% have executed a Big Data project 5% have or are creating a Big Data strategy Source: "State of the Industry Research Series: Big Data in Retail" from Edgell Knowledge Network (E KN) 6
AS and A level specifications in business must encourage students to: develop an enthusiasm for studying business gain an holistic understanding of business in a range of contexts develop a critical understanding of organisations and their ability to meet society’s needs and wants understand that business behaviour can be studied from a range of perspectives generate enterprising and .
