Big Data Use-Cases
across industries

Georg Polzer
+41 79 308 97 23 – georg.polzer@teralytics.ch
Why Big Data Use-Cases? Why not?

 ‣  Today: Need to sell Big Data to Business
     ‣  What can it do for us? , answer with use-case
     ‣  Need to calculate business case
 ‣  Tomorrow: Data first, Business case later
     ‣  Requires data infrastructure (built today)
Pain vs. Lust
 ‣  Use data to solve immediate business
   pain

     ‣  E.g. Manufacturing line inefficient,
          return-rate high, computation takes
          too much time

 ‣  Explorative analysis for data-driven
   innovation

     ‣  You don t know what you will find
     ‣  Drivers: Curiosity and fun
Iterative vs. Disruptive

 ‣  Improve search results vs. self-driving car
 ‣  Try 5 different products simultaneously, collect data
   rigorously, fail fast, double down on success

     ‣  Long tail, fail in order to succeed (mindset!)
     ‣  Natural selection, try to push convergence rate
     ‣  Data over experience (reality changes fast)
     ‣  A/B Testing
Maturity levels of Big Data
 ‣  Level 1: Empower existing business models
     ‣  Understand customer, better service, better products
 ‣  Level 2: Enable data-driven, disruptive innovation
     ‣  Understand past better, start predicting future
 ‣  Level 3: Create data-driven business models
     ‣  Bank sells data about customer-group buying habits to
        retailers, advertisers

     ‣  Mobile network operator predicts traffic jams
Recap: Big Data Use-Cases
         Industry      Data Processing          Advanced Analytics

Web                 Clickstream Sessionization Social Network Analysis

Media               Clickstream Sessionization Content Optimization

 elco               Mediation                  Network Analytics

Retail              Data Factory               Loyalty & Promo

 inancial           Trade Reconciliation       Fraud Analysis

 ederal             SIGINT                     Entity Analysis

 ioinformatics      Genome Mapping             Sequence Analysis
Recap: Use-Case Patterns

 ‣  Data Processing
     ‣  Data enrichment, data transformation
     ‣  Part of ETL Pipeline
 ‣  Complex Analysis
     ‣  Network Analysis (who interacts with whom,
       flow of goods)

     ‣  Correlation, Classification, Clustering
Big Data Use-Cases Checklist

 ‣  Thinking hard does not bring solution (Intelligence
   vs. Statistics)

 ‣  Large amounts of data available for analysis
     ‣  Think out of the box: where do we get data
        from outside the company to fill data gap?

 ‣  Difficult question
     ‣  How much ice-cream did we sell?    vs How
        much ice-cream will we sell next week?
Caveats

 ‣  Targeted advertisement by browser Cookies
   threatened by EU legislation

 ‣  Judging reliability of external data sources in certain
   use-cases crucial (e.g. reputational risk
   assessments)

 ‣  Data privacy barriers very high in Europe
What about you?

Big Data Use-Cases across industries (Georg Polzer, Teralytics)

  • 1.
    Big Data Use-Cases acrossindustries Georg Polzer +41 79 308 97 23 – georg.polzer@teralytics.ch
  • 2.
    Why Big DataUse-Cases? Why not? ‣  Today: Need to sell Big Data to Business ‣  What can it do for us? , answer with use-case ‣  Need to calculate business case ‣  Tomorrow: Data first, Business case later ‣  Requires data infrastructure (built today)
  • 3.
    Pain vs. Lust ‣  Use data to solve immediate business pain ‣  E.g. Manufacturing line inefficient, return-rate high, computation takes too much time ‣  Explorative analysis for data-driven innovation ‣  You don t know what you will find ‣  Drivers: Curiosity and fun
  • 4.
    Iterative vs. Disruptive ‣  Improve search results vs. self-driving car ‣  Try 5 different products simultaneously, collect data rigorously, fail fast, double down on success ‣  Long tail, fail in order to succeed (mindset!) ‣  Natural selection, try to push convergence rate ‣  Data over experience (reality changes fast) ‣  A/B Testing
  • 5.
    Maturity levels ofBig Data ‣  Level 1: Empower existing business models ‣  Understand customer, better service, better products ‣  Level 2: Enable data-driven, disruptive innovation ‣  Understand past better, start predicting future ‣  Level 3: Create data-driven business models ‣  Bank sells data about customer-group buying habits to retailers, advertisers ‣  Mobile network operator predicts traffic jams
  • 6.
    Recap: Big DataUse-Cases Industry Data Processing Advanced Analytics Web Clickstream Sessionization Social Network Analysis Media Clickstream Sessionization Content Optimization elco Mediation Network Analytics Retail Data Factory Loyalty & Promo inancial Trade Reconciliation Fraud Analysis ederal SIGINT Entity Analysis ioinformatics Genome Mapping Sequence Analysis
  • 7.
    Recap: Use-Case Patterns ‣  Data Processing ‣  Data enrichment, data transformation ‣  Part of ETL Pipeline ‣  Complex Analysis ‣  Network Analysis (who interacts with whom, flow of goods) ‣  Correlation, Classification, Clustering
  • 8.
    Big Data Use-CasesChecklist ‣  Thinking hard does not bring solution (Intelligence vs. Statistics) ‣  Large amounts of data available for analysis ‣  Think out of the box: where do we get data from outside the company to fill data gap? ‣  Difficult question ‣  How much ice-cream did we sell? vs How much ice-cream will we sell next week?
  • 9.
    Caveats ‣  Targetedadvertisement by browser Cookies threatened by EU legislation ‣  Judging reliability of external data sources in certain use-cases crucial (e.g. reputational risk assessments) ‣  Data privacy barriers very high in Europe
  • 10.