Auswertung von Kunden- und Betriebsdaten

Data Mining for CRM and Risk Managementkdlabs AGwww.kdlabs.comDr. Jörg-Uwe Kietz

http://www.kdlabs.com/

© January 2003, kd labs ag, analytical CRM and data mining

Overview

Knowledge Discovery @ kdlabs Knowledge Discovery Services (KDS) for CRM

Some cases:– Credit Risk Scoring (to get the good customer)– Customer Life-cycle Analysis for Cross-Selling (to earn more from them)– Customer & Employee Satisfaction (to keep them)

Knowledge Discovery Applications (KDA)– Detecting Money Laundering Activities

Summary


About kdlabs

kdlabs AG was founded in July 2000 to deliver services and to develop applications in the area of Knowledge Discovery Services (KD) and Knowledge Discovery Application (KDA).

kdlabs core competence is KD and KDA. In addition, kdlabs staff has extensive experience in complementary fields, such as Marketing and Marketing Research, CRM and e-CRM, Data Warehousing and Application Integration.

While kdlabs is vendor-independent, it is part of a strong partner network when it comes to the implementation of complete KDA- and CRM-solutions.


incr

ease

pro

fitab

ility

optim

ise

risk

Marketing & CRM Applications• customer acquisition• cross- and up-selling• churn prediction & retention• customer satisfaction modelling• employee satisfaction modelling

Website Applications• website behaviour analysis• website development• dynamic personalisation

Credit Risk Applications• credit risk scoring• credit risk monitoring

Fraud Detection Applications• fraud detection• money laundering detection

Basic Applications(e.g. data quality assessment, profitability analysis, customer segmentation)

Focus on application fields


Overview




Summary


Evolution of customer relation over time

Val

ue o

f cus

tom

er re

latio

n

Three simple business goals of CRM

CustomerAcquisition

Acquire the „right“ customers with high potential

valueCustomer

Development

Cross- and up-sell by offering

the right products at the right time

CustomerRetention

Retain profitable customers and increase their

long-term value

KD for CRM


The CRM Project

Return

Inve

stm

ents „Big Bang“

Need for a managed evolution

„Flop“

„No Go“


Overview




Summary


Credit-Risk Scoring

Not only banks have a credit-risk, everyone who accepts to deliver services or goods before he receives payment has one.

Models predicting credit-risk can be build from previous cases. These models have to be applied on-line at the point-of-sales They can be build off-line, but should be rebuild periodically kdlabs build a predictive model for an advertisement supplier

predicting non-payers with accuracy of 79% on the test set This model is integrated into the customers point-of-sales

application.


How does it work

Customer,Customer history,Background data

AnalysisData

Offline Credit-Risk Modeling

Online Credit-Risk Prediction

CurrentTransaction

Data

Mining

Data P

reprocesssing ScoredTransaction

EnrichedTransaction


Point-of-Sales credit-risk-scoring

The model acquired by KD is used to provide immediate feedback into the sales process


Overview




Summary


Customer Life-cycle Analysis

The Business Problem:– Life insurances are profitable products of insurance-companies– Only a small percentage of all customers own one Cross-sell life-insurances to customers

The Approach:– Transform the “state-description”

Customer C (a ….) owns product P (with …) since X– into a life-cycle description

A Customer C started with P1 at X1, X2 years later she bought P2, …– “Data Mine” typical states of customers when they bought a life-insurance– Rate customers based on their similarity with such prototypical states– Advertise a life insurance to the N highest rated customers

Resulted in one of the most successful campaigns


Buying Age Distribution of Insurances

0%

5%

10%

15%

20%

25%

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100

Age

Cus

tom

ers

Car

Home

Life


Overview




Summary


Customer & Employee Satisfaction

Business Goals• Employee and customer satisfaction are prerequisites for loyal

behavior and customer relation, and by that influence profitability.• Only detailed knowledge about the influence-factors of customer and

employee satisfaction enable effective actions to increase satisfaction, loyalty and therefore profitability.

Therefore:• New Management Methods like Total Quality Management (TQM) or

Balanced Scorecard (BSC) contain the assessment of customer and employee satisfaction.


Customer & Employee Satisfaction

Data Preparation• clean Values• outlier detection• missing values• ...

Causal modeling• factor analysis• business needs

Data Completion

Impact Analysis• Linear Regression• LISREL• PLS• ...

Segmentation• by region• by business process• by division• ...

Result Presentation• Report• Workshop•

The Knowledge Discovery Process Before KD starts


kdimpact Portfolio


kdimpact graph and charts


Overview




Summary


Detecting Money Laundering Activities

The Business Problem Size of worldwide money laundering per year US$ 590-1‘500 billion Over 95% of delinquency sum still undiscovered Criminal potential obvious since September 11, 2001; top-priority for

countering the financing of terrorism Significant damage of reputation and high fines for involved financial

institutions and managers FATF (financial action task force) demands for stronger regulations in

affiliated countries Governments strengthen anti-money laundering laws and regulations Effective Money Laundering detection by bank‘s helps to protect the

secrecy of banking


Examples of what has to be detected transactions from/to uncooperative countries or exposed persons unusual high cash deposits high level of activity on accounts that are generally little used withdrawal of assets shortly after they were credited to the account many payments from different persons to one account repeated credits just under the limit fast flow of a high volume of money through an account

and many more ... e.g. have a look at: – FIU‘s in action: 100 cases from the Egmont Group– Yearly report of the Swiss MROS

Detecting Money Laundering Activities

http://www.gfsc.guernseyci.com/documents/fiu_in_action_full.pdf

http://internet.bap.admin.ch/d/aktuell/berichte/mros-2001_d.pdf


Data analysis

1 2patterns

3

Sel

f-his

tory

Pee

r gro

ups

Link

Ana

lysi

s

rules

expe

rts,

regu

latio

ns

names

Bla

cklis

ts,

PE

P‘s

, etc

.

Overview

bank´stransactions& customers

data

!Alertdelivery

Admin Client

User Interfaces

WorkflowClient

datarepository

externaldata

Helmuth Fuchs

Farblich zeigen, dass kdprevent die Gesamtlösung bietet (mit Schnittstellen)



data

Data analysis

1 2namesrules patterns

3

time

serie

s

outli

ers

etc.

Blac

klis

ts,

PEP

‘s, e

tc.

expe

rts,

regu

latio

ns

!Alertdelivery

Admin Client

User Interfaces

WorkflowClient

datarepository

externaldata

Data collection and data integration

data retrieval from any operational sources and transformation into consistent database

basic data sources: customer profile (CIF and related data), customer view (linked CIF’s), accounts, transactions and many other behavior oriented data

automatic and robust loading process with supportive error correction business-related description of data structure for the ease of use

simultaneously in other parts of the system statistical and quality measures for focused tuning of all technical

processes



data

Data analysis


3

time

serie

s

outli

ers

etc.

Blac

klis

ts,

PEP

‘s, e

tc.

expe

rts,

regu

latio

ns

!Alertdelivery

Admin Client

User Interfaces

WorkflowClient

datarepository

externaldata

Data analysis: three core detection techniques

Data analysis

21patterns

3names

Bla

cklis

ts,

PE

P‘s

, etc

.rules

expe

rts,

regu

latio

ns

Sel

f-his

tory

Pee

r gro

ups

Link

Ana

lysi

s

specific rulesand thresholdslaw, regulations,domain expertise

TvT Complianceinternal experts

unusual patternsand profileshistorical comparison,peer comparison, linkanalysis, etc.

suspicious namesand actorsprimary sources

specialized tools

OFAC

internal lists

Eurospider

Logica

Factiva

World-Check

Etc.

http://www.eurospider.com/index.htm

http://www.logicacmg.com/



data

Data analysis


3

time

serie

s

outli

ers

etc.

Blac

klis

ts,

PEP

‘s, e

tc.

expe

rts,

regu

latio

ns

!Alertdelivery

Admin Client

User Interfaces

WorkflowClient

datarepository

externaldata

Data analysis: detecting unusual patterns / profiles

Pattern discovery 1: self history• e.g. unusual activity in an account history based on

multidimensional time series analysis and comparison time series analysis and comparison

Pattern discovery 2: peer groups• e.g. unusual behavior compared to peer group based on

natural clusters and/or pre-defined segments clustering, segmentation and outlier detection

Pattern discovery 3: link analysis• e.g. similarities in different accounts based on connected/linked

transactions that are not otherwise expected to occur Pattern detection and matching


Overview




Summary


Summary

Knowledge Discovery Services – need a clear business value– data are available everywhere– adequate data preparation is the key success factor– think big, start small search for the most valuable and realistic sub-problems to solve

Knowledge Discovery Applications – enable application, that are not possible without KD– The applications provide a fixed context that

enables the automatisation of the KD-process

Auswertung von Kunden- und Betriebsdaten

Documents

Transcript of Auswertung von Kunden- und Betriebsdaten