Best Practices In Predictive

Analytics
Keeping things simple

June 10, 2014

Ajay Gopikrishnan
Lead Architect Analytics & Bigdata
What is Predictive Analytics (PA)?

Definition
Predictive Analytics is the application of statistical
techniques and BI technologies to uncover
relationships and patters from within large volumes of
data that can be used to predict behavior or events of
interest
Source: TDWI.org

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 2
Predictive Analytics in action

Predictive ! Predictive analytics (PA) is used to address churn

Analytics is more
forward looking
compared to ! PA is used to predict the likelihood of response to a mailer
regular BI we
use past events
to anticipate the
future! ! PA is used to predict the risk of default on a credit card

! PA is used to predict the average time to failure for a particular industrial

heavy machine

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 3
Why is PA not widely deployed?

! It is usually complex and calls for a combination of skills

! The value generated is often under-rated
! Software is expensive
! Dependency on good quality data
! PA is often taken up more as an experiment and not core to function

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 4
Best practices in Predictive Analytics

Data &
Technology

Structured
Unstructured /
semi-
structured

Organization Best Practices

Process

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 5
Best practices in Predictive Analytics

It is important to measure the RoI of PA projects

simply because organizational resources are
ploughed into action

Impact
A Canadian bank
uses PA to
Marketing Mailers are sent increase
campaign
response rates by
Applications are denied / downgraded 600%,
Credit Cards cut customer
acquisition costs
in half, and boost
campaign ROI by
100%. TDWI.ORG

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 6
Best practices in Predictive Analytics

It is more important to evaluate PA projects by

using a set of Business metrics rather than
analytical metrics

Analytical Metrics Business Metrics

No one gets a
! R-square ! Response rate raise or a bonus
based on R-
! Lift ! Gross sales square or lift !
! ROC Curve ! Net Profit - Build business
metrics into the
Analytical Plan

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 7
Best practices in Predictive Analytics

The goal of PA is to isolate the variables from

amongst a large set that can best explain the
event or behavior of interest; normally EDW
tables cannot be used as such

Data jobs Process jobs

Be prepared to
! Merges & joins ! Exploratory analysis spend a good of
deal of your
! Transformations using central tendency project time (75%
measures sometimes) on
! Data Quality data management
! Detect outliers

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 8
Best practices in Predictive Analytics

PA models have a learning curve so be

prepared to stay invested over time to improve
performance and reap benefits

You are trying to

model human
behaviour so do
not expect a
silver bullet
expect
incremental
improvements in
current level of
organizational
performance

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 9
Best practices in Predictive Analytics

Complex Better

Better is usually
A neural network is not necessarily better than measured in
terms of business
Linear Regression if the basic assumptions of metrics
Linear regression are being met in the given
business problem

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 10
Best practices in Predictive Analytics

PA projects can be outsourced to an expert

agency after suitable due diligence based on
resource availability and comparative cost-
benefit analysis with respect to in-sourcing

Outsourcing models

! Leading organizations are known to set up captive analytics

service centers in remote locations where skills are available
! BPOs/KPOs are known to undertake process elements in a
PA project

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 11
Best practices in Predictive Analytics

PA project teams comprise multiple skills most

successful teams branch out as Information
Management as a bridge between business
and IT

PA teams
There is a
! Business analyst substantial
shortage of Data
! Quantitative expert Science skills in
the market;
! Tools expert
Companies are
tying up with
academic
institutions for
PA programs

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 12
Best practices in Predictive Analytics

PA Projects require executive sponsorship; it is

better to adopt a top-down approach, that
originated from the business even if they are
small projects to begin with

Let it not be
reduced to a
research effort
by a PA
enthusiast
within a line
function

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 13
Best practices in Predictive Analytics

PA needs to exploit the Enterprise Data

Warehouse and enabling technology like In-
database and In-memory for scale, faster
throughout and a more comprehensive
approach (think enterprise PA!)
Example technology
Cost of
! SAS + Teradata deployment is
high; hence
! SPSS on Netezza need to measure
RoI of PA
! Analytical Sandboxes projects

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 14
Best practices in Predictive Analytics

There is no best software for Predictive

Analytics; the major contribution is efficiency; it
is up to the user to design the project, define
KPIs, evaluate candidate models and choose
the best model appropriate for the problem

PA + Big Data + Cloud

PA is a
! PA applications are deployable on the cloud combination of
both art and
! PA vendors are now building compatibility with Hadoop and science!
related technology to handle unconventional data types

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 15
The PA Maturity Curve

Business Information Management

Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 16
PA can create business value for the organization

Thank you!

Ajay Gopikrishnan
Lead Architect Big Data & Analytics
[email protected]

Reference: TDWI 2010 paper by Thomas Rathburn - 10 mistakes to avoid in predictive analytics
Business Information Management
Best Practices in Predictive Analytics | June 2014
Copyright 2014 Capgemini. All rights reserved. 17