Business process based analytics

Business Analytics
Business Process and Analytics

KAIST
박준성 교수

2012. 12. 13

Copyright © 2012. Dr. June Sung Park. All rights reserved.

Big Data

What is the most important part
of the term big data?

big

data

both

neither

2

Big Data

What is the most important part
of the term big data?

big

data

both

neither

What organizations do with big data is what is most important. The
analysis your organization does against big data combined with the
actions that are taken to improve your business are what matters.
Analytics only produces business value if it is incorporated into business
processes, enabling business managers and users to act upon the
findings to improve organizational performance.

Bill Franks, Taming the Big Data Tidal Wave, Wiley, 2012.
3

Enterprise Analytics Capability

Using analytics in mainstream business activities is one of
the effective habits of a successful organization.

4

Case Study: Amcor

Company
Global packing company in Australia
with 20+K employees and $7B revenue
in 2008.

Challenge
CEO initiated “Vale Plus” approach
requiring measurement of “pocket
margin” of 1M products down to the
invoice level.

5

Case Study: Amcor

Project
 Phase I:
• Built a data warehouse over a year
consolidating data from 32 apps.
• Biggest challenge was standardizing
and cleansing data.
 Phase II:
• Built analytics and visualization over
a month using an easy-to-use tool.
• Business people with knowledge of
processes helped fit the BI app into
Amcor’s processes.
• Rolled out the BI app in 4 stages,
conducting usability tests, and
training users by their business
managers.

6

Case Study: Amcor

Result Lessons Learned
 500 users adopted it for daily use Link BI projects with strategic initiatives.
in 3 months. Align BI output with corporate KPIs.
 Incentives introduced that are Implement BI within business processes.
based on pocket margins. Have business people, not IT people,
 Corporate gross margin determine BI use cases, i.e., where and
improved. how they would use the BI app in process
execution.
Take enough time to consolidate silo
data into a single version of standardized
and cleansed data.
Pick a tool easy for rapid deployment and
easy for users.
Have business managers train users.

7

Pattern-Based Business Strategy

Technology Market

Your
Company
Supplier Competition

Enterprises should listen to signals and understand when the signals
are patterns that require adaptation.
Listening requires enterprises to combine traditional data sources with
new sources of data.
Gartner, Pattern-based strategy: compete in the new economy using
Gartner’s business pattern framework, Sept. 14, 2005.
8

Pattern-Based Business Strategy

Collective

Defined Creative
Case Management
Business Process Social Media
Database / Data Mart / Warehouse Enterprise Mobility
Service-Oriented Architecture Big Data Analytics
Process Orchestration Complex Event Processing
Event-Driven Architecture
Anticipated Exceptions Anticipated Exceptions

Enterprises should focus their investments on a balanced diversity of
business activities in the defined, creative, collective and exceptions
categories that enable them to innovate and respond to change of patterns.

9

Case Study: Investments in Big Data Analytics

TXU Energy installed smart electric meters in customer homes and read the
meter every 15 minutes. Based on an analysis of the metering data, it applies
dynamic pricing to shape demand curve during peak hours. This eliminates the
need for adding power generating capacity, saving millions of dollars for the
company and saving customer expenditures as well.

T-Mobile USA has integrated data across multiple IT systems to combine
customer transaction and interactions data in order to better predict customer
defections. By leveraging social media data along with transaction data from
CRM and billing systems, T-Mobile USA has been able to “cut customer
defections in half in a single quarter”.

US Xpress collects about a thousand data elements ranging from fuel usage to
tire condition to truck engine operations to GPS information, and uses this data
for optimal fleet management and to drive productivity saving millions of dollars
in operating costs.

10

Pace-Layered IT Strategy

Case Management
Explorative Apps Social Media
Enterprise Mobility
Big Data Analytics
Exploitative Apps Complex Event Processing
Event-Driven Architecture

Stable Digital Foundation Business Process
Database / Data Mart / Warehouse
Service-Oriented Architecture
Process Orchestration

Applications move across layers as they mature, or as the business
process shifts from experimental to well-established to industry standard.
You, however, cannot innovate on an unstable foundation.
Many apps for both disruptive and sustaining innovations should be
based on processes and data in the stable foundation

Gartner, Accelerating innovation by adopting a pace-layered application strategy, Jan. 9. 2012.
11

Evolution of Enterprise IT

Matured enterprise architecture is today based on standardized
and integrated processes and data, and service-oriented
Mobile Cloud
architecture of apps. Computing
(2010-2015)
Technical Debt Payoff
(2005-2010)
E-Business

Process-Orchestrated
(1995-2005)

Mobile + Social + Cloud

Cloud Services
Client/Server

+ Big Data
Computing

Process Management
(1990-1995)
IT Dark Age

IT Modernization

SOA-Based
Online (1980-90)

Process Integration
Computing

Standardization

EA-Based
(1970-80)
Batch
Computing IT
(1950-1970)
Reengineering
Process

Stable Digital Foundation

J. W. Ross, P. Weill and D. C. Robertson, Enterprise Architecture as Strategy, HBS Press, 2006.
12

Stable Digital Foundation

Business process management (BPM) is an ideal technology for agile development
of explorative, exploitative and core apps for an enterprise.
SOA embodies the middle-out architecture where business processes can be
reengineered in flight to quickly implement new business use cases reusing core
business services.
Business service repository and data federation layers virtualize and synchronize
physical apps and data to provide an integrated and standardized foundation.

Composite Apps

Business Process Composition
Business Service Repository
Metadata-Based Data Federation
Physical Apps
Physical Data Sources

Gartner, EIM reference architecture: an essential building block for
enterprise information management, Sept. 14, 2005.
13

Enterprise Architecture

Business Architecture BPM  ACM
Business-IT Alignment

Application Architecture SOA  EDA

Data Architecture MDM  Big Data

Technical Architecture TRM  Virtualization

EA is a strategy planning process ensuring business-IT alignment across
the enterprise using the architectural approach.
Matured EA employs BPM, SOA and MDM disciplines to enable quick
alignment between business use cases and app delivery by reusing
common master data and core services.

14

Enterprise Architecture

Strategy Plan Demand Plan

As Is To Be

BA

AA

IT Asset DA Investment
Mgmt Plan
TA

Transformation

Project Portfolio Mgmt

15

Business Process Management

BPMN 2.0
Design
Graphical modeling,
process simulation,
business rules BPEL4WS, BPEL4P
Implement
Code generation

Execute BPMS
Automation, workflow and
integration
Monitor
Business activity monitoring,
automated process discovery and
dashboards
Optimize
Analyze and dynamically adjust
business processes and rules

16

Business Process Modeling

Enterprise
Content Mgmt
Data Analytics

Data modeling is designing the intended use of data.
Process and data modeling cannot be done separately.

17

Business Process Reengineering

Adaptive
Case Management
Social Collaboration

Process innovation is often enabled by redesigning the flow of
information.

18

Adaptive, Intelligent and Social BPM
Analytics, social network and adaptive case management are integrated into BPM
for performance monitoring and reporting, forecasting, scenario modeling,
complex decisions, planning, real-time situation recognition, immediate next
action recommendation, etc.
Enterprises need business process and performance management maturity that
enables cross-functional accountability and top-down/bottom-up information
flows.

Enterprise
Content Mgmt
Data Analytics

Adaptive
Case Social
Management Collaboration
Forrester, Forrester wave: dynamic case management, Jan. 31, 2011.
19

Adaptive, Intelligent and Social BPM
Integration of analytics into operational processes—which contrasts with past
approaches that separated analytical work from transactional work—
empowers the workforce to make better and faster contextualized decisions
in order to guide work toward optimal outcome, and its impact is immediately
apparent to business people because it changes the way they do their jobs.

http://bps.opentext.com/resources/ot_bps_OT-Process360_ds.pdf
20

BPM Maturity Model
Enterprises usually cannot skip maturity levels.
Enterprises should develop a long-term roadmap to improve their
maturity level, based on the current state assessment and the readiness
check for the next immediate actions.
SOA

EA
BPR

iBPMS

BSC

Gartner, ITScore overview for business process management, Sept. 17, 2010.
21

Advanced BPM Initiatives
Tomorrow's business operations require
integration of real-time intelligence.
Process is the unifying construct for intelligent
operations.
Integration of BPM and automated analytics into
SOA-based iBPM is an important business
evolution underway.

Gartner, Business process management key initiative overview, July 22, 2011.
22

iBPMS
Talend provides open source solutions for data integration, data profiling, data
cleansing, master data management, enterprise service bus, Hadoop connection,
cloud enablement, and BPM.
Using Talend solutions, you can load data from multiple sources into a master
data hub as a SoR, apply the data quality tool to resolve data conflicts, and
provide clean data services for automated decisions in business processes or for
business workers whose workflow is orchestrated by BPM.

23

iBPMS
iBPMS has 10 core
components:
Orchestration engine
for processes and
cases
Model-driven
composition
Human-driven
workflow
Content-driven
workflow
Connectivity of
process to resources
Active analytics
On-demand analytics
Business rule
management
Process repository
BPMS administration

24

Service-Oriented Enterprise Architecture

Portal

Business
Process

Business Service

SaaS
Component

Metadata
Service

Data Mart / Warehouse

Database

Big Data

25

SOA Implementation using BPM Suite

BPEL Process
Process Redesign using BPMN Process KPI Definition Process Simulation Implementation

Service BPM UI and Monitoring Service Integration Test
Specification Implementation Realization and Execution

26

Linking BPM to Analytics based on SOA: SAP Netweaver

BPM-specific BI content in
InfoCube (star schema)

OLAP data

Query on InfoCube

Result
in WSDL

Dashboard rendering
data from BPM

27

Enterprise Information Management

Information
governance and
metadata
management is
critical to any
initiative that
uses data to
drive
improvements to
business
outcome.

28

Enterprise Information Management

29

Enterprise Information Management Initiative
Through 2015, 85% of Fortune 500 organizations will be unable to exploit
big data for competitive advantage.

1 Explore fundamental technology trends, such as big data, mobile, social
media, cloud computing, and how they reinforce each other to offer
opportunities and risks.
2 Plan based on business strategy and enterprise architecture.

3 Model business requirements and detail specification for solution delivery.

4 Choose technologies and vendor/service providers.

5 Implement, test and release the solution iteratively, seeking user feedback.

6 Operate the solution, measure performance, revise the solution and refine
governance processes.

Gartner, Information innovation: innovation key Initiative overview, Apr. 27, 2012.
30

Analytics Framework
Analytic apps can work with any kind of data, including transactions, events,
unstructured contents, website data, social networks, and Internet of things
(machines, sensors).
Increase
Analytics, however, should resolve management challenges first. analytical skills
Embed analytics into the Establish corporate of centralized
business process and workflow. performance metrics. analytic team as
well as self-
service analysts
within business
Attract units.
corporate execs
to participate.
Ensure data
quality and
Find use cases consistency.
and justify
business cases.
Build
requirement
Create engineering
organization competency.
culture of
valuing fact- Consumerize Balance between standardization and diversification,
based decisions. through mobile custom-design and packaged apps, on-premise and
delivery. cloud, SQL and NoSQL, storage and in-memory
Gartner, Analytics key Initiative overview, July 22, 2011.
31

Analytics Maturity Model
Enterprises usually cannot skip maturity levels.
Enterprises should develop a long-term roadmap to improve their
maturity level, based on the current state assessment and the
readiness check for the next immediate actions.

Gartner, ITScore overview for business intelligence and performance management, Sept. 17, 2010.
32

Analytics Roadmap Planning
Enterprises should assess the current level of maturity using a analytics
framework, find areas of weakness and opportunities for improvement,
set up a long-term roadmap to raise the maturity level, follow the EA
process to determine and execute short-term improvement initiatives,
and put in a continuous improvement program.

Data Consistency and Quality

Culture of Analytic
Fact-Based Decision Competencies

Requirement
Process
Engineering
and Metrics
Methodology

Exec Commitment and Governance
33

Analytics Lifecycle

Acquire data Organize data

ETL or
ELT Data platform
Data source (DB, DW,
Hadoop)
Set requirements Select and build
and hypotheses models
Analyze
Take BPM Analytics data
action
for insight
Embed into
Extract rules
operation
BRM

Make decision

34

Analytics Methodology: IBM and Capgemini

35

Analytics Methodology: IBM

36

Analytics Requirement Metamodel
Big data needs big process. (Forrester Research)
Big data without a process context and a compelling use case for a
specific user class is like a Maserati without an engine.
Big data with proven values will become structured.
Process Model Use Case Model UX Model
Business Process Actor Use Case Persona
Rule Actor

I/O Info Process Event Communication
Activity Association

Use Case User Task

Information Model
Service
Data Use Case User Task
Scenario Scenario
Dictionary
Analytics

User Concept
Glossary Data Model Map

37

Analytics Requirement Engineering for SOA

Enterprise
Business Architecture
Strategy
Conceptual
Process Model
UX Conceptual
Model Data Model Business
Use Case Conceptual
Req’ts
Model Service Model

Executable
Process Model Software
Logical Data Req’ts
Schema
UI Use Case
Design Scenario

Analytics

Test Service
Case Specification

38

Analytics Requirement Engineering for SOA

Design
Model

Portal
UX

UI
Business

Case
Test
Process
Scenario
Business
Case
Use

Service

SaaS
Component
Process

Process
Model

Exec

Metadata
Service

Data Mart /
Service
Service

Warehouse
Model

Spec

Database
Schema

Big Data
Model

Data
Data

Service-Oriented Architecture

39

Analytics Requirement Engineering for SOA: IBM

Service Service Process
Use Case Model Process Model Specification Implementation Orchestration

Industry
Reference Model

Data Model

IBM, Building service-oriented solutions with IBM industry models and
Rational software development platform, 2007.
40

Analytics Requirement Engineering for SOA: Capgemini

41

Case Study: PayPal

Company
Global e-commerce business allowing payments and money transfers to be
made through the Internet.
Role of Global Business Analytics Team
 Managing Down: Ensure connection between the analysis they do and the
actions the company takes. Work closely together with business people
for right questions and right interpretation of findings.
 Managing Up: Establish themselves as thought partners, not data
providers, to the executive, and translate analytical insights into actionable
recommendations. Veronika
Belokhvostova, Head
Analytics Team Members of Global Business
Business analysts with a mix of technical and business skills. Most having Analytics at PayPal
MBAs in addition to data analysis skills.
Project Examples
Analysis of customer behaviors and interactions for improving products and
marketing, analysis of the impact of website redesign, analysis of the effect
of promotional pricing, diagnosis of of revenue leakages, analysis of the
impact of risk management policies on customers, etc.

Renee Ferguson, Mining data at PayPal to guide business strategy (Interview with
Veronika Belokhvostova), MIT Sloan Management Review, Sept. 2012.
42

Process-Driven Big Data Analytics Initiative

 Big data analytics requires a data-
savvy business strategy to achieve
competitive advantage.
 Keep the process transparent; it is
key to successful big data projects.
 Educate process owners about
potential big data opportunities
now readily available through start-
small, cost-effective analytics tools
and techniques.
 The value delivered from an
investment in big data analytics
must be visible and measureable.

43

Process-Driven Big Data Analytics Initiative

 Use low-cost, open-source tools in
early pilots to demonstrate the
feasibility of big data projects.
 Explore the increasing number of
public datasets now available through
open APIs.
 Produce a resource plan that identifies
big data skill gaps. Look for business-
savvy analysts (especially data
scientists) and analytics-savvy
business leaders who can work
together to find what business should
do based on analytic results and then
do it.
 Assess resource needs for information
infrastructure and identify technical
gaps when supporting big data
solutions.

44

Data Scientist

Business Use Cases

Analytics Apps

Analytics Common Services

RT-OLAP Analytic Algorithms Visualization
e.g. BigQuery e.g. Greenplum e.g. Pentaho

In-Memory Data Data Models ETL
e.g. GridGain e.g. NoSQL, RDB e.g. Kettle

Basic Data Transformation
e.g. Map Reduce, Pig, Hive, Sqoop, Lucene

File System NoSQL DB
e.g. HDFS e.g. Hbase

(In-Memory) Stream Processing
e.g. Flume, Avro

Distributed Agents

Thomas Davenport and D. Patil, Data scientist: the sexiest job of the
21st century, Harvard Business Review Oct. 2012.
45

Case Study: Sears

Company
American chain of department stores
Challenge
 Decided to generate greater value from
the huge amounts of customer, product
and promotion data collected from its
stores.
 Took 8 weeks, due to highly fragmented
databases and data warehouses, to
generate personalized promotions, at
which point many of them were no
longer optimal.

Andrew McAfee and Erik Brynjolfsson, Big data: the management
revolution, Harvard Business Review, Oct. 2012.
46

Case Study: Sears

Solution
 Set up a Hadoop cluster in 2010,
and used it to store incoming data
from its stores and to hold data
from existing data warehouses.
 Conducted analyses directly on the
cluster, with the processing time
reduced from 8 to 1 week, and still
dropping.
 Got help from Cloudera initially,
but over time internal IT and
analysts became comfortable with
the new tools and methods.

47

Business process based analytics

More Related Content

What's hot (20)

Similar to Business process based analytics (20)

More from Korea Advanced Institute of Science and Technology (20)

Business process based analytics