Professor of Business Analytics and Big Data - bzb.be congres - Philippe Baecke... · Professor of Business Analytics and Big Data ... Visiting Prof. at UCD & Kaplan Business ...

© Prof. dr. Philippe Baecke

PROF DR. PHILIPPE BAECKE

Professor of Business Analytics and Big Data� Prof. at Vlerick Business School� Affiliated Prof. at Ghent University� Visiting Prof. at Trinity College (Dublin)� Visiting Prof. at UCD & Kaplan Business School (Hong Kong)� Visiting Prof. at Université de Namur

Research� Business Analytics & Big Data� (electronic) Customer Relationship Management� Digital Marketing� Spatial & network analysis

Ghent Campus, Office [email protected].: +3292109228

https://be.linkedin.com/in/philippebaecke

THE VALUE OF ANALYTICS IN FINANCIAL SERVICES

PROF. DR. PHILIPPE BAECKE


DigitalFinancial Services

Impact


IMPACT FROM DIGITAL

Margins

8%

12%

18%

23% 22%

44%

0%

10%

20%

30%

40%

50%

<100000 EUR 100000 -

200000 EUR

200000 -

300000 EUR

300000 -

500000 EUR

500000 -

750000 EUR

>750000 EUR

Consolidation of agents and brokers

2014

2016

(source: Benthurst & co – 2016)


IMPACT FROM DIGITAL

Cost reduction

Customer experience


BIG DATA


BIG DATA

Google trends: Big Data

PhDEssays on Data Augmentation:

The Value of Additional Information Creating Business Value with Big Data

Data Driven Marketing


BIG DATA STRATEGY


DATA COLLECTION

Trans-actions

E-mail

www

Social

sensors

Mobile

callDepartment 1

Department 2

Department 3

Company

Customer

sales

…

Touch points Business

Data silo

Data silo

Data silo

Data silo

Data silo


DATA COLLECTION

Trans-actions

E-mail

www

Social

sensors

Mobile

callDepartment 1

Department 2

Department 3

Company

Customer

sales

…

Touch points Business

Single customer view


ANALYSE - DESCRIPTIVE

Data collection / IT infrastructure

Descriptive Analytics



DISCOVERING

Descriptive

Predictive


ANALYSE - PREDICTIVE




PredictiveAnalytics



Available data Predictions

(unknown)

T

Today


Today T-1Independent

variables

T-1

TodayDependent

Variable

(known)


Data mining

technique

Crash



Probability ?

ChurnCrash

Fraud

Purchase

…

Default


BIG DATA


BIG DATA



PredictiveAnalytics

Web/clickstream

Social

Mobilesensors

…

Text

Big Data


BIG DATA - TELEMATICS

Claims

Baecke P. & Bocca L. (2017) - The value of vehicle telematics data in insurance risk selection processes

� Claim history (bonus malus, years without claim, …)

� Customer specific (age, driving experience, …)

� Car specific (kilowatt, brand, …)

Predictive performance

Random

Perfect

+

Driving behaviour

+50%


BIG DATA - TELEMATICS

Descriptive driver feedback


BIG DATA - SOCIAL

Tobback E. & Martens D. - Credit scoring for microfinance using Facebook data

Network of Friends Pseudo-Network of Friends

vs

A B

DC


BIG DATA - SOCIAL


29 traditional variables


Random

Perfect

Default

+� Friends� Likes� Comments� …


BIG DATA - SOCIAL


29 traditional variables


Random

Perfect

+50%Default

+� Friends� Likes� Comments� …


PRESCRIPTIVE ANALYTICS


PredictiveAnalytics

PrescriptiveAnalytics



PRESCRIPTIVE ANALYTICS

Human input replaced by business rules or optimisation algorithms


ROBO-ADVISER

Risk assessment(based on 8 questions)

� Automatic portfolio suggestion (mainly ETFs)� Automatic dividend reinvestments� Automatic tax loss harvesting� Convenient dashboard� …


ARTIFICIAL INTELLIGENCE



PredictiveAnalytics

Big Data

Web/clickstream

Social

Mobilesensors

…

Text

PrescriptiveAnalytics

Artificial intelligence



1997Chess

Kasparov vs

IBM Deepblue

2011JeopadryKen, Brad

vs IBM Watson


COGNITIVE COMPUTINGDescriptive Analytics

PredictiveAnalytics

2007�2011: Jeopardy – Human vs machine



PredictiveAnalytics

Human

Create corpus

Q&A training



PredictiveAnalytics

US Cities



PredictiveAnalytics

Not perfect yet …


CUSTOMER CAREDescriptive Analytics

PredictiveAnalytics

Phase 1:

Phase 2:



1997Chess

Kasparov vs

IBM Deepblue

2011JeopadryKen, Brad

vs IBM Watson

2016Go

Lee Sedolvs

Google Alpha Go


DEEP LEARNING

+ able to detect very complex patterns

- Needs a lot of data observations to be trained well

100 billion neurons

Neuron

Axon


DEEP LEARNING

0 – no damage

0.25 – light damage

0.5 – moderate damage

0.75 – heavy damage

1 – total destruction

Karoon Rashedi Nia (2017) - Automatic Building Damage Assessment Using Deep Learning and Ground-Level Image Data


DEEP LEARNING

Facial analytics to predict life expectancy


DEEP LEARNING


BIG DATA


Algorithms Algorithms


ADOPTION BARRIERS � ENABLERS

CultureCulturePeoplePeople

Customer, privacy

Customer, privacy

Tools, systems,processes

Tools, systems,processes


CUSTOMER & PRIVACY

Prof. dr. Philippe BaeckeAssociate Professor of Marketing

be.linkedin.com/in/philippebaecke

[email protected]

Thank you

Vlerick Business SchoolReep 19000 Gent, Belgiumwww.vlerick.com

Programme director:� Creating Business Value with Big Data� Strategic Data Driven Marketing

Interest to collaborate?

Professor of Business Analytics and Big Data - bzb.be congres - Philippe Baecke... · Professor of Business Analytics and Big Data ... Visiting Prof. at UCD & Kaplan Business ...

Documents