SlideShare une entreprise Scribd logo
1  sur  20
Deutsche Bank
Matthias Meul

Supporting the branch network using
predictive analytics
Predicting first investment behaviour
12Dec2013
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

The next 40 minutes
—
—
—
—
—

Introduction to Deutsche Bank & casestudy
The First Investment model
Commercial implementation & results
Lessons learned
Q&A

Deutsche Bank
12/18/2013 11:56:11 AM

Matthias Meul
First Investment model
2010 DB Blue template

Introduction
Model

Results

Questions

1
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Deutsche Bank in Belgium
—
—
—
—
—
—
—

> 315 000 clients
+-715 employees (of which +-300 with client contact roles)
33 Financial Centers
3 Contact Centers
Over 21 billion € AUM
―Open architecture‖, mandates as of 100€
Retail – DbPersonal – Private

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

2
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Casestudy overview
— Clients with investments are treated very well (maturities, downgrades,…)
— The objective is to convert retail savers-only to investors
A significant number of clients doesn’t have and never had
investments!

Deutsche Bank
12/18/2013 11:56:11 AM

Matthias Meul
First Investment model
2010 DB Blue template

3
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Dependent variable
— Will the client make a first investment in a strategic product in the year to
come? Yes/No
— Timeframe?
T3: Start of the period in which the behavior will take place

Historical data taken into account to
make the prediction
T2: The moment when we want to make the prediction
T4: End of the period in which the behavior will take place

Period to execute commercial initiatives based on predictions

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

4
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Modeling technique
— Linear regression
— Dependent variable between -∞ and +∞
— Easy to explain:
— However, the problem of identifying First Investors is binary
— Made a FI = 1
— Did not make a FI = 0
— The desired outcome of the model are probabilities (between 0%
and 100%)
We have to perform a transformation
— Logistic regression
— Model the log odds of the event (=dependent variable), which will be
transformed into probabilities
Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

5
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Variables retained (high level) using stepwise regression
—
—
—
—
—
—

Current AUM & variations
AUM in the past
Education level
Current level of loyalty towards Deutsche Bank (~advanced algorithm)
Length of relation
…

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

6
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Evaluation
— Check model performance… Wait until time goes by…?
T3: Start of the period in which the behavior will take place

Historical data taken into account to
make the prediction
T2: The moment when we want to make the prediction

30jun2012

T4: End of the period in which the behavior will take place

Period to execute commercial initiatives based on predictions

31jul2012

Use the model to score,
based on 28feb2013

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

31jan2013

Monitor model performance

31mar2013

30sep2013

7
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Technical performance measures
— Lift
— AUC

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

8
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Technical performance measures: Lift
— Lift

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

9
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Technical performance measures: AUC
— Area Under the receiver operating Curve
— The probability that a classifier will rank a randomly chosen positive
instance higher than a randomly chosen negative one
A random ―not‖ first investor (score =
0.005%)

Customers

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

A random first investor
(score = 10%)

When I compare the scores of the FI
and the non-FI, the FI has 70% chance
(AUC = 0.70) of having a higher
predicted probability
of churning than the non-FI

10
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Operational implementation of the First Investment model
— Recurrent Tasks, monthly refresh
— Event-driven (e.g. AUM evolution)No ice-cold calling
— always at least 1 of 3 potential commercial triggers in Task comment

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

11
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Results of the first roll-out

+-20% of treated tasks results in a meeting!

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

12
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

―Educate‖ the branch network

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

13
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Results of the first roll-out (purchases until end of september)

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

14
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Results of the first roll-out (purchases one month after task injection)

Sales until October 19th taken into account because of
*tasks being treated as of September 19th
*time between ordering and booking an investment product

Deutsche Bank
12/18/2013 11:56:12 AM

Matthias Meul
First Investment model
2010 DB Blue template

15
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Is it necessary to contact these clients?
Won’t they invest ―by themselves‖?
Do they really need the extra trigger?...
Same characteristics as targeted clients & compare!

Deutsche Bank
12/18/2013 11:56:13 AM

Matthias Meul
First Investment model
2010 DB Blue template

16
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Results of the First Investment model
First Investors 2012 vs 2013 (31nov)

Investors_per_month_2012
Investors_per_month_2013
YTDInvestors_2012
Total2012

YTDInvestors_2013

Jan

Feb

Deutsche Bank
12/18/2013 11:56:11 AM

Mar

Apr

May

Jun

Matthias Meul
First Investment model
2010 DB Blue template

Jul

Aug

Sep

Oct

Nov

Dec

17
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Lessons Learned
— Do not limit yourself to the existing KPIs. Do they match the business
problem you’re trying to solve?
— Assure the commercial implementation of the model, put the scores to use
& add tangible customer leads.
— Assure the followup of the model & recurrently report results.
— Keep it simple & understandable!

Deutsche Bank
12/18/2013 11:56:13 AM

Matthias Meul
First Investment model
2010 DB Blue template

18
Introduction

Model

Results

Questions

First Investment
Predictive analytics in Investments

Thank you!

Deutsche Bank
12/18/2013 11:56:13 AM

Matthias Meul
First Investment model
2010 DB Blue template

19

Contenu connexe

Similaire à Case Study: Supporting the branch network using predictive analytics: Predicting First Investment behavior

GUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCH
GUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCHGUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCH
GUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCHGuido Perboli
 
BSA Public Sessions | Public Subsidies with CP Wackernagel
BSA Public Sessions | Public Subsidies with CP WackernagelBSA Public Sessions | Public Subsidies with CP Wackernagel
BSA Public Sessions | Public Subsidies with CP WackernagelBerlinStartupAcademy
 
IT Consultants as Change Agents in Digital Transformation Initiatives
IT Consultants as Change Agents in Digital Transformation Initiatives IT Consultants as Change Agents in Digital Transformation Initiatives
IT Consultants as Change Agents in Digital Transformation Initiatives Nicolai Krüger
 
Improving Customer Performance as a Guide for Business Model Innovation -
Improving Customer Performance as a Guide for Business Model Innovation -Improving Customer Performance as a Guide for Business Model Innovation -
Improving Customer Performance as a Guide for Business Model Innovation -Wilhelm Graupner, Ph.D.
 
Horizon 2020 SME Instrument and Eurostars Proposal Development
Horizon 2020 SME Instrument and Eurostars Proposal DevelopmentHorizon 2020 SME Instrument and Eurostars Proposal Development
Horizon 2020 SME Instrument and Eurostars Proposal DevelopmentBusiness West
 
Primer on Business Modeling Canvas
Primer on Business Modeling CanvasPrimer on Business Modeling Canvas
Primer on Business Modeling CanvasHeimo Hänninen
 
Data-Driven Value Generation. Is it Possible?
Data-Driven Value Generation. Is it Possible?Data-Driven Value Generation. Is it Possible?
Data-Driven Value Generation. Is it Possible?M2M Alliance e.V.
 
Cigo 2008
Cigo 2008Cigo 2008
Cigo 2008Dutch2K
 
TeamProsource PMI Benelux 2010, Become the 360° project manager - teaser
TeamProsource PMI Benelux 2010, Become the 360° project manager - teaserTeamProsource PMI Benelux 2010, Become the 360° project manager - teaser
TeamProsource PMI Benelux 2010, Become the 360° project manager - teaserTeamProsource
 
2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led
2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led
2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc ledSirris
 
The Busisness Case for Digital Workplace Projects
The Busisness Case for Digital Workplace ProjectsThe Busisness Case for Digital Workplace Projects
The Busisness Case for Digital Workplace ProjectsStephan Schillerwein
 
Business Meets IT Keynote 24/6/2014 William Visterin
Business Meets IT Keynote 24/6/2014 William VisterinBusiness Meets IT Keynote 24/6/2014 William Visterin
Business Meets IT Keynote 24/6/2014 William VisterinWilliam Visterin
 
Christel visser curriculum eng
Christel visser curriculum engChristel visser curriculum eng
Christel visser curriculum engChristel Visser
 

Similaire à Case Study: Supporting the branch network using predictive analytics: Predicting First Investment behavior (20)

GUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCH
GUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCHGUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCH
GUEST: A LEAN METHODOLOGY FOR MANAGEMENT SCIENCE AND OPERATIONS RESEARCH
 
BSA Public Sessions | Public Subsidies with CP Wackernagel
BSA Public Sessions | Public Subsidies with CP WackernagelBSA Public Sessions | Public Subsidies with CP Wackernagel
BSA Public Sessions | Public Subsidies with CP Wackernagel
 
Jerome Fisse C V 1.4
Jerome  Fisse C V 1.4Jerome  Fisse C V 1.4
Jerome Fisse C V 1.4
 
IT Consultants as Change Agents in Digital Transformation Initiatives
IT Consultants as Change Agents in Digital Transformation Initiatives IT Consultants as Change Agents in Digital Transformation Initiatives
IT Consultants as Change Agents in Digital Transformation Initiatives
 
Improving Customer Performance as a Guide for Business Model Innovation -
Improving Customer Performance as a Guide for Business Model Innovation -Improving Customer Performance as a Guide for Business Model Innovation -
Improving Customer Performance as a Guide for Business Model Innovation -
 
Horizon 2020 SME Instrument and Eurostars Proposal Development
Horizon 2020 SME Instrument and Eurostars Proposal DevelopmentHorizon 2020 SME Instrument and Eurostars Proposal Development
Horizon 2020 SME Instrument and Eurostars Proposal Development
 
Primer on Business Modeling Canvas
Primer on Business Modeling CanvasPrimer on Business Modeling Canvas
Primer on Business Modeling Canvas
 
BBC - Better Business Cases - Foundation
BBC - Better Business Cases - FoundationBBC - Better Business Cases - Foundation
BBC - Better Business Cases - Foundation
 
Graphical CV
Graphical CVGraphical CV
Graphical CV
 
Data-Driven Value Generation. Is it Possible?
Data-Driven Value Generation. Is it Possible?Data-Driven Value Generation. Is it Possible?
Data-Driven Value Generation. Is it Possible?
 
Cigo 2008
Cigo 2008Cigo 2008
Cigo 2008
 
TeamProsource PMI Benelux 2010, Become the 360° project manager - teaser
TeamProsource PMI Benelux 2010, Become the 360° project manager - teaserTeamProsource PMI Benelux 2010, Become the 360° project manager - teaser
TeamProsource PMI Benelux 2010, Become the 360° project manager - teaser
 
Eig guidelines
Eig guidelinesEig guidelines
Eig guidelines
 
2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led
2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led
2015 02-11-eco-innovate-in-lighting-12 13 & 14 key takeaways feedback cyc led
 
The Busisness Case for Digital Workplace Projects
The Busisness Case for Digital Workplace ProjectsThe Busisness Case for Digital Workplace Projects
The Busisness Case for Digital Workplace Projects
 
Agile BI success factors
Agile BI success factorsAgile BI success factors
Agile BI success factors
 
Project Evaluation
Project EvaluationProject Evaluation
Project Evaluation
 
Presentation3
Presentation3Presentation3
Presentation3
 
Business Meets IT Keynote 24/6/2014 William Visterin
Business Meets IT Keynote 24/6/2014 William VisterinBusiness Meets IT Keynote 24/6/2014 William Visterin
Business Meets IT Keynote 24/6/2014 William Visterin
 
Christel visser curriculum eng
Christel visser curriculum engChristel visser curriculum eng
Christel visser curriculum eng
 

Plus de BAQMaR

Sam Wouters - Blockchain and the big data/market research industry
Sam Wouters - Blockchain and the big data/market research industrySam Wouters - Blockchain and the big data/market research industry
Sam Wouters - Blockchain and the big data/market research industryBAQMaR
 
Maarten Verschuere - A perfect storm: when market research and data science meet
Maarten Verschuere - A perfect storm: when market research and data science meetMaarten Verschuere - A perfect storm: when market research and data science meet
Maarten Verschuere - A perfect storm: when market research and data science meetBAQMaR
 
Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...
Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...
Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...BAQMaR
 
Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...
Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...
Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...BAQMaR
 
Ludovic Depoortere - Virtual reality meets sensory research
Ludovic Depoortere - Virtual reality meets sensory researchLudovic Depoortere - Virtual reality meets sensory research
Ludovic Depoortere - Virtual reality meets sensory researchBAQMaR
 
Yuri Van Geest - Exponential Organizations
Yuri Van Geest - Exponential OrganizationsYuri Van Geest - Exponential Organizations
Yuri Van Geest - Exponential OrganizationsBAQMaR
 
Denyse Drummond-Dunn - Winning Customer Centricity
Denyse Drummond-Dunn - Winning Customer CentricityDenyse Drummond-Dunn - Winning Customer Centricity
Denyse Drummond-Dunn - Winning Customer CentricityBAQMaR
 
Stijn Geuens - I know what you’ll buy next summer
Stijn Geuens - I know what you’ll buy next summerStijn Geuens - I know what you’ll buy next summer
Stijn Geuens - I know what you’ll buy next summerBAQMaR
 
Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...
Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...
Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...BAQMaR
 
Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...
Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...
Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...BAQMaR
 
Anouk Willems - Turning Insights into Company-wide Memes
Anouk Willems - Turning Insights into Company-wide MemesAnouk Willems - Turning Insights into Company-wide Memes
Anouk Willems - Turning Insights into Company-wide MemesBAQMaR
 
Anouar El Haji - Auctions Speak Louder than Words
Anouar El Haji - Auctions Speak Louder than WordsAnouar El Haji - Auctions Speak Louder than Words
Anouar El Haji - Auctions Speak Louder than WordsBAQMaR
 
Christophe Ovaere - Disrupting the Traditional MR Model
Christophe Ovaere - Disrupting the Traditional MR ModelChristophe Ovaere - Disrupting the Traditional MR Model
Christophe Ovaere - Disrupting the Traditional MR ModelBAQMaR
 
Ray Poynter - Keynote: The Mobile Future of Research & Analytics
Ray Poynter - Keynote: The Mobile Future of Research & AnalyticsRay Poynter - Keynote: The Mobile Future of Research & Analytics
Ray Poynter - Keynote: The Mobile Future of Research & AnalyticsBAQMaR
 
Jon Puleston - Survey Research: The Science of ‘Prediction’
Jon Puleston - Survey Research: The Science of ‘Prediction’Jon Puleston - Survey Research: The Science of ‘Prediction’
Jon Puleston - Survey Research: The Science of ‘Prediction’BAQMaR
 
Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...
Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...
Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...BAQMaR
 
Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...
Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...
Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...BAQMaR
 
Tom De Ruyck - Opening session: Disrupt or be Disrupted
Tom De Ruyck - Opening session: Disrupt or be DisruptedTom De Ruyck - Opening session: Disrupt or be Disrupted
Tom De Ruyck - Opening session: Disrupt or be DisruptedBAQMaR
 
Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...
Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...
Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...BAQMaR
 
Kristof Coussement - The Debate: the Future of (Big) Data Analytics Software
Kristof Coussement - The Debate: the Future of (Big) Data Analytics SoftwareKristof Coussement - The Debate: the Future of (Big) Data Analytics Software
Kristof Coussement - The Debate: the Future of (Big) Data Analytics SoftwareBAQMaR
 

Plus de BAQMaR (20)

Sam Wouters - Blockchain and the big data/market research industry
Sam Wouters - Blockchain and the big data/market research industrySam Wouters - Blockchain and the big data/market research industry
Sam Wouters - Blockchain and the big data/market research industry
 
Maarten Verschuere - A perfect storm: when market research and data science meet
Maarten Verschuere - A perfect storm: when market research and data science meetMaarten Verschuere - A perfect storm: when market research and data science meet
Maarten Verschuere - A perfect storm: when market research and data science meet
 
Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...
Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...
Prof. dr. Vincent F. Hendricks - Online bubbles and the downsides of social m...
 
Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...
Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...
Daphne Fecheyr Lippens - Biomimicry: learning from nature for disruptive inno...
 
Ludovic Depoortere - Virtual reality meets sensory research
Ludovic Depoortere - Virtual reality meets sensory researchLudovic Depoortere - Virtual reality meets sensory research
Ludovic Depoortere - Virtual reality meets sensory research
 
Yuri Van Geest - Exponential Organizations
Yuri Van Geest - Exponential OrganizationsYuri Van Geest - Exponential Organizations
Yuri Van Geest - Exponential Organizations
 
Denyse Drummond-Dunn - Winning Customer Centricity
Denyse Drummond-Dunn - Winning Customer CentricityDenyse Drummond-Dunn - Winning Customer Centricity
Denyse Drummond-Dunn - Winning Customer Centricity
 
Stijn Geuens - I know what you’ll buy next summer
Stijn Geuens - I know what you’ll buy next summerStijn Geuens - I know what you’ll buy next summer
Stijn Geuens - I know what you’ll buy next summer
 
Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...
Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...
Chloé Van Vreckem - Uncovering the true Customer Value by using Survival Anal...
 
Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...
Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...
Andy Petrella - Data Science is changing and you won’t be allowed to claim yo...
 
Anouk Willems - Turning Insights into Company-wide Memes
Anouk Willems - Turning Insights into Company-wide MemesAnouk Willems - Turning Insights into Company-wide Memes
Anouk Willems - Turning Insights into Company-wide Memes
 
Anouar El Haji - Auctions Speak Louder than Words
Anouar El Haji - Auctions Speak Louder than WordsAnouar El Haji - Auctions Speak Louder than Words
Anouar El Haji - Auctions Speak Louder than Words
 
Christophe Ovaere - Disrupting the Traditional MR Model
Christophe Ovaere - Disrupting the Traditional MR ModelChristophe Ovaere - Disrupting the Traditional MR Model
Christophe Ovaere - Disrupting the Traditional MR Model
 
Ray Poynter - Keynote: The Mobile Future of Research & Analytics
Ray Poynter - Keynote: The Mobile Future of Research & AnalyticsRay Poynter - Keynote: The Mobile Future of Research & Analytics
Ray Poynter - Keynote: The Mobile Future of Research & Analytics
 
Jon Puleston - Survey Research: The Science of ‘Prediction’
Jon Puleston - Survey Research: The Science of ‘Prediction’Jon Puleston - Survey Research: The Science of ‘Prediction’
Jon Puleston - Survey Research: The Science of ‘Prediction’
 
Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...
Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...
Filip Maertens - Artificial Intelligence: Building Emotion & Context aware Re...
 
Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...
Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...
Corinne Sandler - Keynote: Wake up or die! Be the only one who does what you ...
 
Tom De Ruyck - Opening session: Disrupt or be Disrupted
Tom De Ruyck - Opening session: Disrupt or be DisruptedTom De Ruyck - Opening session: Disrupt or be Disrupted
Tom De Ruyck - Opening session: Disrupt or be Disrupted
 
Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...
Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...
Wim Hamaekers - Neuro Science: Car Clinics 3.0 – Winner Esomar Effectiveness ...
 
Kristof Coussement - The Debate: the Future of (Big) Data Analytics Software
Kristof Coussement - The Debate: the Future of (Big) Data Analytics SoftwareKristof Coussement - The Debate: the Future of (Big) Data Analytics Software
Kristof Coussement - The Debate: the Future of (Big) Data Analytics Software
 

Dernier

Q3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast SlidesQ3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast SlidesMarketing847413
 
Call Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
Log your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaignLog your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaignHenry Tapper
 
High Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur Escorts
High Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur EscortsHigh Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur Escorts
High Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur Escortsranjana rawat
 
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...shivangimorya083
 
VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130
VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130
VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130Suhani Kapoor
 
Best VIP Call Girls Noida Sector 18 Call Me: 8448380779
Best VIP Call Girls Noida Sector 18 Call Me: 8448380779Best VIP Call Girls Noida Sector 18 Call Me: 8448380779
Best VIP Call Girls Noida Sector 18 Call Me: 8448380779Delhi Call girls
 
Quarter 4- Module 3 Principles of Marketing
Quarter 4- Module 3 Principles of MarketingQuarter 4- Module 3 Principles of Marketing
Quarter 4- Module 3 Principles of MarketingMaristelaRamos12
 
03_Emmanuel Ndiaye_Degroof Petercam.pptx
03_Emmanuel Ndiaye_Degroof Petercam.pptx03_Emmanuel Ndiaye_Degroof Petercam.pptx
03_Emmanuel Ndiaye_Degroof Petercam.pptxFinTech Belgium
 
Pooja 9892124323 : Call Girl in Juhu Escorts Service Free Home Delivery
Pooja 9892124323 : Call Girl in Juhu Escorts Service Free Home DeliveryPooja 9892124323 : Call Girl in Juhu Escorts Service Free Home Delivery
Pooja 9892124323 : Call Girl in Juhu Escorts Service Free Home DeliveryPooja Nehwal
 
CALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual service
CALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual serviceCALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual service
CALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual serviceanilsa9823
 
The Economic History of the U.S. Lecture 20.pdf
The Economic History of the U.S. Lecture 20.pdfThe Economic History of the U.S. Lecture 20.pdf
The Economic History of the U.S. Lecture 20.pdfGale Pooley
 
VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...
VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...
VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...Suhani Kapoor
 
Instant Issue Debit Cards - School Designs
Instant Issue Debit Cards - School DesignsInstant Issue Debit Cards - School Designs
Instant Issue Debit Cards - School Designsegoetzinger
 
Call US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure service
Call US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure serviceCall US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure service
Call US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure servicePooja Nehwal
 
Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )
Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )
Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )Pooja Nehwal
 
Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...
Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...
Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...ssifa0344
 
20240417-Calibre-April-2024-Investor-Presentation.pdf
20240417-Calibre-April-2024-Investor-Presentation.pdf20240417-Calibre-April-2024-Investor-Presentation.pdf
20240417-Calibre-April-2024-Investor-Presentation.pdfAdnet Communications
 
The Economic History of the U.S. Lecture 19.pdf
The Economic History of the U.S. Lecture 19.pdfThe Economic History of the U.S. Lecture 19.pdf
The Economic History of the U.S. Lecture 19.pdfGale Pooley
 
20240429 Calibre April 2024 Investor Presentation.pdf
20240429 Calibre April 2024 Investor Presentation.pdf20240429 Calibre April 2024 Investor Presentation.pdf
20240429 Calibre April 2024 Investor Presentation.pdfAdnet Communications
 

Dernier (20)

Q3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast SlidesQ3 2024 Earnings Conference Call and Webcast Slides
Q3 2024 Earnings Conference Call and Webcast Slides
 
Call Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Koregaon Park Call Me 7737669865 Budget Friendly No Advance Booking
 
Log your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaignLog your LOA pain with Pension Lab's brilliant campaign
Log your LOA pain with Pension Lab's brilliant campaign
 
High Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur Escorts
High Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur EscortsHigh Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur Escorts
High Class Call Girls Nagpur Grishma Call 7001035870 Meet With Nagpur Escorts
 
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
Russian Call Girls In Gtb Nagar (Delhi) 9711199012 💋✔💕😘 Naughty Call Girls Se...
 
VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130
VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130
VIP Call Girls Service Dilsukhnagar Hyderabad Call +91-8250192130
 
Best VIP Call Girls Noida Sector 18 Call Me: 8448380779
Best VIP Call Girls Noida Sector 18 Call Me: 8448380779Best VIP Call Girls Noida Sector 18 Call Me: 8448380779
Best VIP Call Girls Noida Sector 18 Call Me: 8448380779
 
Quarter 4- Module 3 Principles of Marketing
Quarter 4- Module 3 Principles of MarketingQuarter 4- Module 3 Principles of Marketing
Quarter 4- Module 3 Principles of Marketing
 
03_Emmanuel Ndiaye_Degroof Petercam.pptx
03_Emmanuel Ndiaye_Degroof Petercam.pptx03_Emmanuel Ndiaye_Degroof Petercam.pptx
03_Emmanuel Ndiaye_Degroof Petercam.pptx
 
Pooja 9892124323 : Call Girl in Juhu Escorts Service Free Home Delivery
Pooja 9892124323 : Call Girl in Juhu Escorts Service Free Home DeliveryPooja 9892124323 : Call Girl in Juhu Escorts Service Free Home Delivery
Pooja 9892124323 : Call Girl in Juhu Escorts Service Free Home Delivery
 
CALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual service
CALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual serviceCALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual service
CALL ON ➥8923113531 🔝Call Girls Gomti Nagar Lucknow best sexual service
 
The Economic History of the U.S. Lecture 20.pdf
The Economic History of the U.S. Lecture 20.pdfThe Economic History of the U.S. Lecture 20.pdf
The Economic History of the U.S. Lecture 20.pdf
 
VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...
VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...
VIP Call Girls LB Nagar ( Hyderabad ) Phone 8250192130 | ₹5k To 25k With Room...
 
Instant Issue Debit Cards - School Designs
Instant Issue Debit Cards - School DesignsInstant Issue Debit Cards - School Designs
Instant Issue Debit Cards - School Designs
 
Call US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure service
Call US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure serviceCall US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure service
Call US 📞 9892124323 ✅ Kurla Call Girls In Kurla ( Mumbai ) secure service
 
Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )
Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )
Vip Call US 📞 7738631006 ✅Call Girls In Sakinaka ( Mumbai )
 
Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...
Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...
Solution Manual for Financial Accounting, 11th Edition by Robert Libby, Patri...
 
20240417-Calibre-April-2024-Investor-Presentation.pdf
20240417-Calibre-April-2024-Investor-Presentation.pdf20240417-Calibre-April-2024-Investor-Presentation.pdf
20240417-Calibre-April-2024-Investor-Presentation.pdf
 
The Economic History of the U.S. Lecture 19.pdf
The Economic History of the U.S. Lecture 19.pdfThe Economic History of the U.S. Lecture 19.pdf
The Economic History of the U.S. Lecture 19.pdf
 
20240429 Calibre April 2024 Investor Presentation.pdf
20240429 Calibre April 2024 Investor Presentation.pdf20240429 Calibre April 2024 Investor Presentation.pdf
20240429 Calibre April 2024 Investor Presentation.pdf
 

Case Study: Supporting the branch network using predictive analytics: Predicting First Investment behavior

  • 1. Deutsche Bank Matthias Meul Supporting the branch network using predictive analytics Predicting first investment behaviour 12Dec2013
  • 2. Introduction Model Results Questions First Investment Predictive analytics in Investments The next 40 minutes — — — — — Introduction to Deutsche Bank & casestudy The First Investment model Commercial implementation & results Lessons learned Q&A Deutsche Bank 12/18/2013 11:56:11 AM Matthias Meul First Investment model 2010 DB Blue template Introduction Model Results Questions 1
  • 3. Introduction Model Results Questions First Investment Predictive analytics in Investments Deutsche Bank in Belgium — — — — — — — > 315 000 clients +-715 employees (of which +-300 with client contact roles) 33 Financial Centers 3 Contact Centers Over 21 billion € AUM ―Open architecture‖, mandates as of 100€ Retail – DbPersonal – Private Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 2
  • 4. Introduction Model Results Questions First Investment Predictive analytics in Investments Casestudy overview — Clients with investments are treated very well (maturities, downgrades,…) — The objective is to convert retail savers-only to investors A significant number of clients doesn’t have and never had investments! Deutsche Bank 12/18/2013 11:56:11 AM Matthias Meul First Investment model 2010 DB Blue template 3
  • 5. Introduction Model Results Questions First Investment Predictive analytics in Investments Dependent variable — Will the client make a first investment in a strategic product in the year to come? Yes/No — Timeframe? T3: Start of the period in which the behavior will take place Historical data taken into account to make the prediction T2: The moment when we want to make the prediction T4: End of the period in which the behavior will take place Period to execute commercial initiatives based on predictions Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 4
  • 6. Introduction Model Results Questions First Investment Predictive analytics in Investments Modeling technique — Linear regression — Dependent variable between -∞ and +∞ — Easy to explain: — However, the problem of identifying First Investors is binary — Made a FI = 1 — Did not make a FI = 0 — The desired outcome of the model are probabilities (between 0% and 100%) We have to perform a transformation — Logistic regression — Model the log odds of the event (=dependent variable), which will be transformed into probabilities Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 5
  • 7. Introduction Model Results Questions First Investment Predictive analytics in Investments Variables retained (high level) using stepwise regression — — — — — — Current AUM & variations AUM in the past Education level Current level of loyalty towards Deutsche Bank (~advanced algorithm) Length of relation … Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 6
  • 8. Introduction Model Results Questions First Investment Predictive analytics in Investments Evaluation — Check model performance… Wait until time goes by…? T3: Start of the period in which the behavior will take place Historical data taken into account to make the prediction T2: The moment when we want to make the prediction 30jun2012 T4: End of the period in which the behavior will take place Period to execute commercial initiatives based on predictions 31jul2012 Use the model to score, based on 28feb2013 Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 31jan2013 Monitor model performance 31mar2013 30sep2013 7
  • 9. Introduction Model Results Questions First Investment Predictive analytics in Investments Technical performance measures — Lift — AUC Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 8
  • 10. Introduction Model Results Questions First Investment Predictive analytics in Investments Technical performance measures: Lift — Lift Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 9
  • 11. Introduction Model Results Questions First Investment Predictive analytics in Investments Technical performance measures: AUC — Area Under the receiver operating Curve — The probability that a classifier will rank a randomly chosen positive instance higher than a randomly chosen negative one A random ―not‖ first investor (score = 0.005%) Customers Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template A random first investor (score = 10%) When I compare the scores of the FI and the non-FI, the FI has 70% chance (AUC = 0.70) of having a higher predicted probability of churning than the non-FI 10
  • 12. Introduction Model Results Questions First Investment Predictive analytics in Investments Operational implementation of the First Investment model — Recurrent Tasks, monthly refresh — Event-driven (e.g. AUM evolution)No ice-cold calling — always at least 1 of 3 potential commercial triggers in Task comment Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 11
  • 13. Introduction Model Results Questions First Investment Predictive analytics in Investments Results of the first roll-out +-20% of treated tasks results in a meeting! Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 12
  • 14. Introduction Model Results Questions First Investment Predictive analytics in Investments ―Educate‖ the branch network Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 13
  • 15. Introduction Model Results Questions First Investment Predictive analytics in Investments Results of the first roll-out (purchases until end of september) Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 14
  • 16. Introduction Model Results Questions First Investment Predictive analytics in Investments Results of the first roll-out (purchases one month after task injection) Sales until October 19th taken into account because of *tasks being treated as of September 19th *time between ordering and booking an investment product Deutsche Bank 12/18/2013 11:56:12 AM Matthias Meul First Investment model 2010 DB Blue template 15
  • 17. Introduction Model Results Questions First Investment Predictive analytics in Investments Is it necessary to contact these clients? Won’t they invest ―by themselves‖? Do they really need the extra trigger?... Same characteristics as targeted clients & compare! Deutsche Bank 12/18/2013 11:56:13 AM Matthias Meul First Investment model 2010 DB Blue template 16
  • 18. Introduction Model Results Questions First Investment Predictive analytics in Investments Results of the First Investment model First Investors 2012 vs 2013 (31nov) Investors_per_month_2012 Investors_per_month_2013 YTDInvestors_2012 Total2012 YTDInvestors_2013 Jan Feb Deutsche Bank 12/18/2013 11:56:11 AM Mar Apr May Jun Matthias Meul First Investment model 2010 DB Blue template Jul Aug Sep Oct Nov Dec 17
  • 19. Introduction Model Results Questions First Investment Predictive analytics in Investments Lessons Learned — Do not limit yourself to the existing KPIs. Do they match the business problem you’re trying to solve? — Assure the commercial implementation of the model, put the scores to use & add tangible customer leads. — Assure the followup of the model & recurrently report results. — Keep it simple & understandable! Deutsche Bank 12/18/2013 11:56:13 AM Matthias Meul First Investment model 2010 DB Blue template 18
  • 20. Introduction Model Results Questions First Investment Predictive analytics in Investments Thank you! Deutsche Bank 12/18/2013 11:56:13 AM Matthias Meul First Investment model 2010 DB Blue template 19

Notes de l'éditeur

  1. I‘mgoingto talkabouthowwebackedupthecontactstrategy @DB withanalytics & predictivemodelingThe mainfocus in thispresentation will be on the First Investment model.I‘mgoingtobeveryhands on, talk aboutissueswefacedandhowwe dealt withthem. Pleasekeep in mindthattheorganizationdoes not havemuchexperiencewithpredictivemodels, so theaimistokeeptheanalyticspractice simple andunderstandableto all stakeholders!Commercial implementation & results: thisisgoingtobeabouthowweputanalyticstowork, andtheresultswegetfromthem.I‘ll do a smallrecap on thelessonslearnednearthe end ofmy talk, andthere will beroomforsome Q&A.
  2. Open architecture: Billboard showingnamesoffundmanagersthatsays „youdon‘thavethischoicewithyourotherbank“ „Not real bankers“This „capacitymanagement“ isonepartoftheIntroductionthatwealwayshavetokeep in mind…Mission of CFU team (Costumerfollowup)+ NO predictivemodelingnoradvancedanalytics 3 yearsago
  3. Add: “Analyse behavior 6 monthsafterfirstinvestmentfor “naturallyconvertedcustomers” in the past: higheraum, closer link with the bank, higherup-sellpotential, more investments,lowerchurn,…“Andthewaywewanttosolvethisisusing „Analytics“~predictivemodeling
  4. The firstquestionyou must askwhenyou’redeveloping a predictive model is: what is itthat I want to predict?Don’ttakethis topic of defining the timeframe lightly! It’s important that the customerbehavioryou’retrying to predictallignswellwith the intended commercialuse of youranalytical model. Likesomany models, we startedoffwith the idea of wanting to predict the behavior over the nextyear, butafterthoroughanalysis we sawthat the events in a 6 month timeframe had analmostidentical incident rate.+ Leavesome time for commercial actions to anticipate the forecastedbehaviourMODEL (“probablyonlyonewith a beautiful lady in hispresentation”)+ Carefulnot to miss commercial opportunities (~commercial implementation: useactualmaturities)
  5. The goalistosolvetheequationto „p“, becausethisistheprobabilityofmaking a firstinvestment
  6. I am not goingtospendtomuch time on themethodsused, splittingintotrainingandvalidation sample, ifyouareinterested in thisyoucan talk tome after thispresentation.But, after modeling, wecameupwiththislistofpredictor variables toinfluence First Investment behavior. Now, how well doesour model perform?
  7. Once we have the model, we can check its performance onrealbehavior (apart of coursefrom the performance on the validationor test set) …Waituntil time goesby? No, of coursenot, youcanalreadyuse the model to score clients in the past and model their performance. Just becarfeulnot to use the same timeframe as the oneyouused to build the model! Ifyou do thiscorrectly, youcanalreadyperform and out of sample out of time test
  8. I’mnotgoing to focus toomuchon “technical performance measureshere, butI’mgoing to talk a bit onhow we put these scores to use!I’mgoing to go over these 2 veryquickly.Also, this is notrocketscience, this was the first model developed @db, sointerpretability was prioritisedaboveachieving the highestpossibletechnical performance. Just soeveryone is on the same page, we now have predictions per customerthatindicate the probability of thatclientmaking a firststrategicinvestment.
  9. Whatdid we do before? (excel, high potentiallists per branch)Recurrenttasks (~explainwhat a task is!), monthlyrefresh”Keep the water fresh”, nolonger “the taskshaven’t been treatedyet, nonewones!”
  10. proactivelycontacting & helpingclientswiththeirfirstinvestmentisnice!On theimportanceofcorrectlylabelingandtreatingtheirtasks!
  11. proactivelycontacting & helpingclientswiththeirfirstinvestmentisnice!On theimportanceofcorrectlylabelingandtreatingtheirtasks!
  12. Task vsNotask: weare not speakingaboutthe same clients, hencethenextslideisneededtobereallyabletojudgeperformanceofthetrigger!
  13. proactivelycontacting & helpingclientswiththeirfirstinvestmentisnice!
  14. Youcan present AUC, lift curves,… all you want, butit’s a slidelikethisthatreallycapturesmanegement’sattention, and truly shows the addedvalue of implementinganalytics in your business processes.1) End of 2012: objectiveachieved, curve levelsoff (~Incentivecaused bias?)2) Results of anentireyear was alreadyachievedmid-september3) February: No tasks
  15. 1. THINK OUT OF THE BOX!Thosearethekeylessonswetrytokeep in mind, andthankstothisWe‘renolongerconsidered a costcenter, we‘renow a profitcenter!Thankyouforyourattention!