Forecasting & Big Data Analytics - JPK Group€¦ · BI and Advanced Analytics are only part of the equation & the CFO has a critical role to play Data tion BI 2nd Gen BI Advanced

JPK

Gro

up

June 14-15 • Boston, MA

Forecasting & Big Data AnalyticsUsing Big Data to Enhance Forecasting and Planning

June 14, 8:30am

View presentation online at:

http://www.jpkgroupsummits.com/attendee3

Paul Legutko – Senior Manager, Enterprise

Intelligence/Digital Analytics at Ernst and Young

Business Forecasting and Analytics Forum

Paul Legutko is a leading expert in digital measurement and analytics and a senior manager

in EY’s Analytics practice. Over the last twelve years as a digital analytics consultant, Paul

has seen the field evolve from raw server log files, web analytics, mobile and social

measurement, and into today’s world of Big Data integration. Working with clients from

many industries – particularly Financial Services, Media & Entertainment, and Pharma --

Paul has pioneered the application of new solutions for understanding customer digital

behavior, digital marketing effectiveness, and how to integrate offline and online data.

Page 1 Forecasting & Big Data

Forecasting & Big

Data Analytics

Using Big Data to Enhance

Forecasting and Planning

with Paul Legutko


But do you know what big data is?

You know what forecasting is…


Most people think of it like this…

► It sounds plausible, but doesn’t it make big data sound like

just more of the same?

Big data “Not so big” data

Volume Terabytes/petabytes Megabytes/gigabytes

Variety Unstructured (text, voice, video) Structured/relational

Velocity Data in motion (streaming) Data at rest

Veracity Untrusted/uncleansed Trusted/cleansed


Traditional data and the Four V’s

Traditional data

(however large, fast, inaccurate or various)

Account

Date Address Name Purchases

Status Opt-in SpendLast

Activity

It doesn’t matter

how large this

gets

It doesn’t matter

how many sources

you have

It doesn’t matter

how fast these

change

Traditional database, reporting and analysis tools can handle it


But knock things down a level …

Big data

(however small, at rest, accurate or singular)

Account

Call OrderDirect

ContactWeb Visit

And you

introduce three

critical new

dimensions

Sequence

The order of events matters

Time

The time between events matters

Pattern

The pattern of events matters


The Sequence of events matters

This set of data

Forecast

SaleCallSaw ad Web visit

is completely different than this

Forecast an

Annoyed

Customer

CallWeb visitSale Saw ad


The time between events matters

These events in time

Forecast

Sale

CallSaw ad Web visit

Yield a different forecast than these events in time

May 3rd May 4th May 7th

Forecast

Unclear

Saw ad

Jan. 2nd

Web visit

Mar. 3rd

Call

July 12th


And the pattern of events matters

This consumption pattern

WeekendWeekend Weekday

Has the same average but a different forecast

High Use Low Use High Use

Weekday

Low Use

Weekend

Forecast

WeekendWeekend Weekday

Med. Use Med. Use Med. Use

Weekday

Med. Use

Forecast

Unclear


Traditionally, we’ve handled this by creating sub-forecasts

This consumption pattern

Weekend

High Use

Weekday

Low Use

Weekend

High Use

Weekday

Low Use

Weekend

Forecast

Weekday

Forecast


But that doesn’t work when patterns are complex or individual

$130

June

$142

July

$145

August

$158

Sep.

$167

Oct.

$187

Nov.

C

1ACCELERATING

$130

June

$82

July

$145

August

$25

Sep.

$135

Oct.

$15

Nov.

C

2 OSCILLATING

$130

June

$150

July

$145

August

$18

Sep.

$25

Oct.

$20

Nov.

C

3 SEASONAL

NO SINGLE AVERAGING STRATEGY WILL YIELD THE RIGHT ANSWER


It’s true in every paradigm case of big data

► Big Data shifts analytics to patterns in detail-level data.

Read meter

once a quarterUtilit

ies

Read meter

every 10

minutes

Aggregate

stream

Analyze the

flow of usage

by day/time

NewOld

Mass market

Reta

il

Cross channel

customer data

Customized

offers

Learn and

respond

Price by risk

‘strata’Auto

Insura

nce

Sensor data

in car

Analyze

driving

patterns

Price risk at

individual level


It matters because when sequence, time and pattern of events are important

Relational

data models

break down

SQL is extremely cumbersome and

almost impossible to use when querying

sequence, time between and patterns

Stats tools

are limited

Traditional analysis methods (like

regression and correlation) don’t work

directly

Integration

strategies

break

Traditional joins become multi-joins and

no longer facilitate useful integration,

even when keys are present


Categorizing Data Sources


So if you didn’t know…now you know

… so how do you forecast with Big Data?

What is big data


Three common big data scenarios

You have unstructured data (text, voice or video) from sources

like Twitter, Call-Center Notes, or published video content

1

You need to apply basic or even complex statistical analysis to

identify known patterns (improving, seasonal, etc.) at the

individual level before aggregating

2

You need to identify patterns in the data before you can

understand what it means

3


Structuring unstructured data

You have unstructured data (text, voice or video) from sources

like Twitter, Call-Center Notes, or published video content

1

Call-Center

Social media

Survey Research

Feedback

Offline research

Once text actions are

categorized they can be

used with traditional

forecasting techniques

Taxonomy

Sentiment

Linguistic Parsing

Machine Learning

Text-Analytics


Applying analytics to each record

You have structured data from sources. You need to apply basic

or even complex statistical analysis to identify known patterns

(improving, seasonal, etc.) at the individual level before

aggregating

2

Records

Clustering Momentum Pattern ID Deep LearningRegression

Spark or

Hadoop

Create an individual forecast for each record. Then aggregate traditionally.


Finding potential patterns in the data

You need to identify patterns in the data before you can

understand what it means

3

Activity 1 Activity 2

Use pattern matching techniques to spot and identify behavioral signatures and then categorize them for aggregation.


The Process has changed…

► Before Big Data

Customer Market Sales

Aggregate

Market Sales

Apply

Analytics

Market Forecast Sales

► Using Big Data


Web Visit Survey Call

Apply

Analytics


Velocity Sentiment Persona

Sale Probability

Aggregate

Market Forecast Sales


What do you need to do?

1

Big data isn’t just having a lot of data – big data problems occur when

the level of meaning resides above the detail level in the sequence,

order or pattern of the events.

2 When that happens, traditional IT techniques break down

3 And so do forecasting techniques

4To make big data useful, you need to add an extra step to your process

where you categorize the data in ways that allow it to be aggregated.

5

There are three common ways to do that: applying text analytics to

structure unstructured text, applying statistical analysis to identify

individual-level trends, and applying pattern-matching to categorize

behavior.

When you’ve done that, you can forecast with big data.


So… How do I do this?

Is the future of “Big Data”, “Big Judgment”?

Forecasting with Big Data


Predictive Analytics and Big Data

Big data and analytics represent a disruptive innovation for

business

Analytics is a strategy, talent, and operations topic, NOT just a

technology and data issue

The value of data – “Big” or “Not So Big” -- lies in the actions

businesses take from the insights obtained…analytics alone has no

value

There are hurdles to overcome to realize value …the hardest ones

are organizational and people related


There are many, many more tools available today than there were three years ago…

► http://mattturck.com/2016/02/01/big-data-landscape/


So the tools are there…But you need the right people to

• Ask the right questions• Interpret the data correctly

…This is Advanced Analytics

Forecasting with Big Data


Case Study

► Organization: The Lydian Empire, 6th century BC, under King Croesus.

► Research Question: “What will happen if Lydia invades Persian territory?”

► Solutions and Vendor Selection: Seven oracles in Africa and Greece.

► Controlled Experiment / POC: “What is King Croesus doing right now?” (answer: cooking tortoise

and lamb stew in a covered bronze cauldron)

► Vendor Selected: The Oracle of Delphi

► Fees: 2 silver & gold mixing bowls, gold statue of Croesus’ baker, 117 gold ingots, a gold lion, and the

Queen’s necklace and girdle.

► First Predictive Output: “If you invade into Persia, you will destroy a mighty empire.”

► Second Predictive Output: “Only if a mule is King of Persia will the operation go badly.”

► Action: Lydia invaded Persian Territory.

► Result: Lydian capital Sardis was sacked. Croesus’ fate is unclear.

Case Study

► Organization:

► Research Question:

► Solutions and Vendor Selection:

► Controlled Experiment / POC:

“________________________________________________________

► Vendor Selected:

► Fees:____________________________________________________________________________

___

► First Predictive Output:

► Second Predictive Output:

► Action:

► Result:

Predictive Analytics is not new


But more data requires more analyticsBI and Advanced Analytics are only part of the equation & the CFO has a critical role to play

Data

Ins

igh

t &

Ac

tio

n

BI

2nd Gen BI

Advanced Analytics

• You want to know how your campaign performed. You want to see how many widgets you produced and sold.

• Organizational weaknesses start appearing and potential opportunity is jeopardized

• Visualization becomes key to telling a story about what’s going on. Basic analytics are providing insights. Dashboards become more interactive.

• But, underlying data platforms and systems start breaking. Organizational silos start cracking – right people, right skills, right org model?

• Harder to ask the question – How is my business performing?

• Advanced analytics proactively providing insights and driving actions.

• Exploratory analytics becomes key combining advanced analytics and visualization to help you make money, save money and save lives.

• Can I be a real time enterprise?

Operational

(What Happened)

Operational + Digital

(What’s Going On)

All Data + Streaming

(What Should I Do)


The Role of the CFO is changing – why?Thinking about BI and Data Analytics

► There’s too much at stake for

traditional roles and silos to stand …

Priority Collaboration What’s at Stake

Managing Cybersecurity Health of the enterprise (Risk)

Creating analytics-driven org Being data driven; grow the Top Line

Establishing information strategy,

architecture

Ability to compete and improve

performance

Data

an

d A

na

lytics

Po

we

rTime

Data Acquisition

Data Visualization

Artificial Narrow Intelligence

Artificial General Intelligence

Today

Data Forecasting

Full Report: http://www.ey.com/GL/en/Issues/Managing-finance/EY-CFO-program-partnering-for-

performance-the-CFO-and-the-CEO-driving-and-enabling


And Organizations must evolve to be data-driven

Full Report: http://www.ey.com/Analytics

Strategy and leadership

► Key findings:

► Analytics is central to business strategy

► Leaders already gain competitive advantage from data and analytics

► Leading enterprises designate leaders to guide initiatives

Data analytics impact index 2015: The ‘Best’ versus the ‘Rest’


The human element of analyticsTechnical capability often exceeds behavioral ‘alignment’

Analytics Maturity Drivers:

► Strategy and Leadership

► Adoption

► Strategic Alignment

► Competitive Differentiation

► Analytics Production

► Technology

► People Analytics Skills

► Data Management

► Analytics Consumption – Organization

► Governance

► Culture

► Change Management

► Analytics Consumption – Individual

► Decision Making

► Insights to Action

► Value Measurement

Technical

Capability Gap

Value

Creation

Danger

Zone

Behavioral

Alignment Gap

Low

Hig

h

► Culture and

mental models

► Organization and

process design

► Learning and

development

► Incentives /

rewards

Behavioral Alignment

(Analytics Consumption)

Low High

► Data quality

► Infrastructure and tools

► Data science

Technical Capability

(Analytics Production)


Example: Analytics Visualization and ReportingThe goal of analytics is to democratize knowledge, not data.

Analysts: Understand

what business

questions reporting is

supposed to answer,

and develop the

metrics and KPI’s that

provides this insight.

Modelers: Combine

historical and current

reporting with

forecasting and what-if

capabilities that

transform them from

reports to tools for

running the business.

Architects: Engineer the automation of reporting derived directly from

data systems, using whatever visualization and BI tools make most sense

for the organization.

Visualization Design Skill Domains

► Strategy

► Industry knowledge

► Operations

Business Analyst Skill Domains

► Strategy

► Accounting

► Finance

► Economics

► Marketing

Modeler Skill Domains

► Data Modeling

► Business Intelligence

► Advanced Analytics

► Statistics

► Visualization

Architect Skill Domains

► Data Management

► Data Warehouse

► BI Platform Implementation

► Data Integration and Connectivity

Designers: Condense

metrics and KPIs into

visually appealing, rich,

easily comprehensible

and interactive

dashboards and reports.

Every Forecast is a Collaborative Effort… And they are all analytics professionals


• Big Data isn’t More Data, it’s Different Data, requiring new methods, tools, and skills to activate.

• Using Big Data Analytics requires• The right tools• The right people• The right culture

Wrapping Up


Thank you

► [email protected]

► +1 617 930 1207

Paul Legutko

EY | Assurance | Tax | Transactions | Advisory

About EY

EY is a global leader in assurance, tax, transaction and advisory

services. The insights and quality services we deliver help build trust

and confidence in the capital markets and in economies the world

over. We develop outstanding leaders who team to deliver on our

promises to all of our stakeholders. In so doing, we play a critical

role in building a better working world for our people, for our

clients and for our communities.

EY refers to the global organization, and may refer to one or more,

of the member firms of Ernst & Young Global Limited, each of

which is a separate legal entity. Ernst & Young Global Limited, a UK

company limited by guarantee, does not provide services to clients.

For more information about our organization, please visit ey.com.

Ernst & Young LLP is a client-serving member firm of Ernst & Young

Global Limited operating in the US.

© 2013 Ernst & Young LLP.

All Rights Reserved.

1309-1135041

Documents

Forecasting & Big Data Analytics - JPK Group€¦ · BI and Advanced Analytics are only part of the equation & the CFO has a critical role to play Data tion BI 2nd Gen BI Advanced