big data sharing experience - jacques wieczorek
TRANSCRIPT
Big Data sharing experience
NRB Mainframe day 2016
Jacques WIECZOREK, Head of BI/Analytics Solutions NRB 26/05/2016
Brussels
J.Wieczorek NRB Mainframe Day 2016
J.Wieczorek NRB Mainframe Day 2016
2. Goals
3. The Data
7. Big Data Hub with the Walloon Region
6. Strategy
8. Challenges
9. Conclusion
5. Definition of the Big Data concept
4. The context of Big Data
1. Introductory statement
101101101
Data is based on :Some basics rules,taxonomy
Crossing the Data increasesthe knowledge
Out of context :Lost of itsMeaning,Lost of its value
Organizational framework of the Data :Who owns it ?Where are the Data ?Who updates it ?Data governance
OmnichannelsData contentCookies, Sessions time,..Mobile behaviorSocial MediaBehavior, groups
ID RegistrationEmails, Gender,Type of channelRegistration,…
E-catalogProductsWhish listsPrice segments…
Some features of the Data : New dimensions
J.Wieczorek NRB Mainframe Day 2016
A product/ a service experience
Customer experience is the new strategic battlefield
CustomerExperience
Use of Technology is exploding
Power of the Mass
The customer decides
J.Wieczorek NRB Mainframe Day 2016
We carry our consumers habits and expectations into any workplace !
B2CB2BG2C
Scope of the customer experience
J.Wieczorek NRB Mainframe Day 2016
Anyway there is a massive volume of both structured and unstructured data
Store
Capture
Store Treatment
Visualization
This huge volume is too big, or moves too fastor exceeds current processing capacity.
J.Wieczorek NRB Mainframe Day 2016
This explosive growth creates new opportunities : Customer experience
Customer centric : serving customers at every touchpoint
Marketing or content strategy to increase loyalty and decrease churn.
Product or services issues
Process gaps
Feature/ service requests
No Revolution
J.Wieczorek NRB Mainframe Day 2016
Big Data : tentative to define the concept of Big Data
BIGDATA
a set of activities that allows deeper and more complex treatment of datasets to generate the data asset into something of value with new enabling technologies
J.Wieczorek NRB Mainframe Day 2016
What is hidden ?..
Label: connectedType :married
Label: connectedType :married
Label: connectedType :married
Label: connectedType : married
Label: shareholder
Are they potential conflict of interests ? :Direct conflicts ?Indirect conflicts
Which company could be involved ?Direct ? Indirect ?
J.Wieczorek NRB Mainframe Day 2016
Customer phases
Discovery ProjectsStrategicfocus
Phase 1
Maintenance IOT
Phase 2 Phase 3 Phase 4
J.Wieczorek NRB Mainframe Day 2016
Other uses cases we are developping
FRAUDDETECTION
PREDICTIVEMAINTENANCE
To determine« fraud patterns » :
manually Automatically Ring Fraud Suspicious
behaviours…..
360° VIEW(Churn/loyalty)
To determine and to plan preventivemaintenance tasks
• To determine shared customercharacteristics :• Loyalty definition,• Churn understanding• Churn trends• Churn models• Recommandation system• Bundling based on
recommandations• ….
J.Wieczorek NRB Mainframe Day 2016
3 basic questions
What is the question you want to answer with big data ?1
2
3
Do you have the data to answer that question ?
If you could answer the question, could you use the answer ?
J.Wieczorek NRB Mainframe Day 2016
Accroître Revenu
RéduireCharges
DuCapital
Accroîtrele Cash Flow
RéduireCoûts
Réduire Capital Utilisé
Réduire Coût du Capital
FondRoulement
Immob
BiensD’équipements
CoûtServices
Prix Ventes
VolumesVentes
FraisAdministratifs
CoûtsCommerciaux
Churn
Fraud
Predictive Maintenance
ROI to be associated with the use case
J.Wieczorek NRB Mainframe Day 2016
Vulnerabilities
To ensure the sustainability of the business by managing the risks of :
Compliance
Security
Privacy
Data Governance
Executive
Drivers
J.Wieczorek NRB Mainframe Day 2016
DATACOMPETITIVENESS JOB CREATIONS COMMERCIALISATION
Allowing all companies of
the Walloon Region to get
access to one single entity
for any Big Data activity
the would like to start with
(R&D projects, Big Data
projects)
Getting the 6
competitiveness clusters
ahead of competition by
applying Big Data
techniques to create value,
to improve operations and
to make faster and more
intelligent decisions.
Securing Data Privacy for
some specific sectors like
the Healthcare sector and
the Space sector
Leading to job creation and
turning R&D outcomes into
production
Vision of the Walloon Region : sustaining the Marshall Plan
J.Wieczorek NRB Mainframe Day 2016
Major components of the Walloon Big Data hub
HardwareSoftware
Researchand Development
Commercialentity
Multi tenant
Dynamic sizing
Large portofolio of Big Data tools
Daas
Agreement with Cenearo
to get some extra capacity
The DGO6 ( one of the seven operational directorates (DGO) of the Walloon Public Service is the key policy-design and implementing body for regional research and innovation policy.
It also supports the
Walloon actors in
business networks.
Legal entity in the form of a SCRL
Provide competitiveness clusters with Big Data support for :
R&D projects,
Big Data projects
J.Wieczorek NRB Mainframe Day 2016
4 use cases have already been identified within the R&D projects
MEDICAL RESEARCHHEALTHCARE AEROSPACE ICT
Real time monitoring
of patients in intensive
care which will supply
a warning system for
doctors
Acceleration of
research results on the
genome and
identification of factors
promoting the
development of
cancers
Preventive
maintenance of test
benches for aircraft
engines
Preventive
maintenance to
anticipate failure to
hardware components
VIS
ION
DIMINUTION
Morbidity
Mortality
Healthcare
Costs
SAVINGS
Alarms
Help to
decision
DATA
INTERPRETATIO
N
BY
CLINICAL
ALGORITHM
LAYER
•Integrator
•Synthesizer
•Pertinent Filters
•Medical AlgorithmsBig Data
BIOCORDER
MOBILE
HEALTHCARE
CLINICAL
ASSISTANT
The DIM 3 Project
J.Wieczorek NRB Mainframe Day 2016
J.Wieczorek NRB Mainframe Day 2016
PIT EORegions !
Partners : Centre Spatial de Liège Ulg, Ecole Royale Militaire, I-Mage Consult, NRB
Services
Dynamic earth observation services by delivering geospatial information based on a regular period of observations.
This will help to determine any change in the the state of vegetation, landslide, mine shaft…
J.Wieczorek NRB Mainframe Day 2016
NRB benefits from 3 business models to sell some Big Data activities :
▪ NRB’s own business environment▪ Research Projects
▪ Commercial entity dedicated to the Walloon Region
Three channels to sell Big Data activities
J.Wieczorek NRB Mainframe Day 2016
StatisticsSporadic
mistakeerror
Recurring Cognitive
Category of challenges
J.Wieczorek NRB Mainframe Day 2016
Confirmation bias
Simpson’s Paradox
mistakeerror
StatisticsSporadic Recurring
Confounding variable
Challenges
J.Wieczorek NRB Mainframe Day 2016
Homer Edward
Same name, different first names, different fates…
Simpson’s Paradox
J.Wieczorek NRB Mainframe Day 2016
Statistical indicators : Survival rate / Period : 10 days
Emergency Services : global indicator of the performances
Identical criteria for :
Hospital A
Hospital B
Additional Information :
Hospital Total Survivors Deaths SurvivalRate
Hospital A 1000 800 200 80 %
Hospital B 1000 900 100 90 %
J.Wieczorek NRB Mainframe Day 2016
Statistical Paradox : a trend that appears in different groups of data disappears or reverses when these groups are combined
ill
Wounded
?
Clarification :• mathematical points of view
• The number of deaths is low within the “ill” population at both Hospitals and Hospital B has more inpatients into this category
• There are more deaths within the “wounded”population and there are more inpatients into this category at the Hospital A/
• It is not correct to add up those categories
• Statistics – medical statistics point of view
• the severity risks factors are not similar for the two categories
• Prior to the measure a set of criteria should have been applied. Therefore this lack of selection degrades the quality of the observation.
.
J.Wieczorek NRB Mainframe Day 2016
Apophenia
Inference fallacy
Confirmation bias
Simpson’ s Paradox
Mistakeerreur
StatisticsSporadic Recurring Cognitive
J.Wieczorek NRB Mainframe Day 2016
listen to your intuition
Theater ticket + parking ticket = 1,1€
The Theater ticket costs one euro more
than the parking ticket
The Parking ticket
?
J.Wieczorek NRB Mainframe Day 2016
Take aways
o Having more data doesn’t susbtitue for thinking hard, recognizinganomalies and exploring deep truths,
o Elaborate your « use case » with a multidisciplinary team
o Don’t fall victim to SOS ( Shiny Object Syndrome ) => do POCS !
o Sherlock Holmes was a data scientist :
« my name is Sherlock Holmes . It is my business to knowwhat other people don’t know ».