Big Data Analytics

Project Management:
Methodologies, Caveats
and Considerations

Tiffani Crawford, PhD

 Speaker and Contact Information
Tiffani Crawford, PhD
408-829-7096
zapquick@aol.com
Big Data Analytics Project Management book by Tiffani Crawford – See Amazon Books at
http://www.amazon.com and search by title and author.

Tiffani Crawford, PhD, builds global Big Data Analytics systems. She has 20 years of high
technology experience with Fortune 500 companies, including Cisco Systems, Cognizant,
Bank of America, VISA/Inovant, BAE Systems, Applied Competitive Technologies, Ditech
Networks/Nuance, Big 4 financial firms, defense contractors and startups. She has worked in
seminal technology development in Big Data, analytics, cloud, networking,
telecommunications, software development, distributed multi-tier applications,
multimedia/digital, geographic information systems, intelligent transport systems, finance,
security, policy systems and structural equation modeling. She is a credited software
developer and published author. She earned her PhD from the University of Southern
California in 2005. She has also earned her Master's, Bachelor's and various technology
certifications. She is a member of PMI with various philanthropic contributions.

2 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Introduction and Contents
Business Case for Big Data Analytics
• What is Big Data Analytics?
• Getting to Business Value
• Opportunities and implications
• Clients

Project Management for Big Data Analytics

• Common misconceptions
• Does Big Data Analytics fit your client’s interest, situation and experience?
• Choosing the best methodology
• Evaluating caveats and risks
• Unique considerations
• Growth and sustainability

Hands-On Exercise
• Create a repeatable model for a client’s Big Data Analytics in a focus area
that inspires you and your team
• Continue this exercise and discussion via social media and online
applications

3 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Big Data – INCOMING!!!!!
• Transactions per Second (TPS) in • Structured
Terabytes (TB) • Semi-Structured
• Storage per Day or Year in • Unstructured
Petabytes (PB) or Exabytes (EB) • Real-time data from
• Hadoop nodes 30+ planned social networks
• Data source systems 10+ • Mobile data
• Dedicated nodes • Machine data
• Multiple data hubs Volume Variety • Imaging data
• Multiple datacenters • External data

• Data streaming Velocity Complexity • Predictive statistics

• Data processing in • Data architecture
Real Time (<5 minutes) • Data pipeline analysis
or Near Real Time (NRT) • Computational models
(<30 minutes) • Insight granularity
• Query and results in <10 seconds • Query and results
• Visualization streaming packaged for usability
• Business reporting and interoperability for
• Application processes and networks analytics applications

4 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Big Data Characteristics

Large quantities of
many data types
• Structured
• Semi-Structured
• Unstructured
• Human
• Machine

5 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Valuable Insights

Getting to Value
• Quantitative
• Qualitative
• Correlation
• Longitudinal
• Social
• Search
• Operational
• Inferential
• Ethnographic
• Interview-Based
• Casual

6 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Valuable Insights
(x + y)/y + <10 sentiment blogs
3y/z

Getting to Value
x/2
• Quantitative
• Qualitative
4xy
• Correlation
• Longitudinal
• Social
• Search
*5z • Operational
• Inferential
*3t if over • Ethnographic
18 years old
• Interview-Based
Data Points • Casual

7 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Big Data Initiative Challenges

Information Lifecycle Specialized Skills

Management Development
Data Centralization Administration
Data Decentralization Data Science

Driving Value Data Architecture

Information Focus Data Governance
Big Data Thinking Metrics and KPIs
Transformation BU Compliance

Application Audience :: Tools

Architecture
Performance
Infrastructure and
Network Usability
Training + Support Visualization

8 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Technology to Business Value

Marketing Customer Network

Visualize Analytics Care Operations Actionable, timely insights
Meaningful • Focused on functional
business business needs
context • Dynamic, interactive apps
in near real-time Business

Analyze Data Streaming Intelligent Huge volumes, streaming data

Value In real time Fusion Analytics Caching • 250B+ Transactions/day
• 100+ TB/day

Consume Analyze, then store

Multiple • 10x to 100x infrastructure
savings (storage, transport)
distributed
• Multiple data types and
Technology
sources
Static Data sources fused and analyzed
Dynamic Data

9 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Business Case for Big Data Analytics
• Applicable to all enterprise applications and clients
• Faster
• High-speed data processing
• 1 big job => many small jobs (reduce complexity)
• Redundancies prevent data loss
• Temporary storage is faster
• Long-term storage is easier to access
• High-speed query and results processing (<10 sec)
• High-speed predictive analytics & responsive models
• Cheaper
• Low cost (2-10x cheaper) up to massive scale
• Technology-agnostic
• Commodity hardware
• In-Memory options
• Interoperates easily with plug-ins
• Open source available

10 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Business Case for Big Data Analytics
• Easier
• Data validation and quality = near-100% accuracy
• Network resilience and 5-9s performance
• Security resilience and management
• Virtualization and automation
• Network traffic management
• Transitory data management
• Data retention policies and management
• Enterprise-wide analytics with quality data
• Query management and reusable code
• Start Small and Practice
• Define business value, use cases and metrics/KPIs
• Open source, cloud or large supplier install base
• Proof of Concept, Proof of Value, Proof of Technology
• Early analytics modeling, Proof of Model
• Pilot, testing, pre-production and production buildouts

11 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Who uses Hadoop?

12 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Hadoop Technology-Agnostic Market

Industry-Specific Big Data Solutions

Business Context
Execution
Components Information Delivery
Analytics & Insights

Processing

Service Integration & Management

Offerings
Data
Consulting Governance

Solution Research &

Big Data Engineering Development
Dimensions
Volume Variety Complexity Velocity

Research & Development

Entrepreneurship
Business Transformation
Market Extension

13 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 4 Pillars – Complete Ecosystems
Hadoop
• Java or alternatives
• SQL
• Components in Ecosystems
NoSQL
• Alternative to SQL databases
• Components in Ecosystems
Hybrid Systems
• Hadoop and/or NoSQL components
• Legacy databases and Enterprise Data Warehouses (EDWs)
• Legacy analytics for understanding what happened
• Install bases from large providers
• Slow ETLs moving data
Data Science
• Analytics for prediction and business transformation
• Data visualization and reporting

14 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Hadoop = MapReduce + HDFS
Master Node Interconnect Data Nodes

Name Data
Data
Node Data Data Data Data Data
Node Node
Node Node Node Node Node
Job Task
Tracker Tracker Task Task Task Task Task
Tracker Tracker Tracker Tracker Tracker
Secondary
(Passive) Node
Data Data Data Data Data
Name Data
Data Node Node Node Node Node
Node
Node Node
Task Task Task Task Task
Job Task Tracker Tracker Tracker Tracker Tracker
Tracker Tracker

Data Data Data Data Data

Node Node Node Node Node

Task Task Task Task Task

Tracker Tracker Tracker Tracker Tracker

15 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Hadoop Ecosystem

MapReduce Software framework for clustered, distributed data processing

HDFS Primary storage system used by Hadoop in a distributed environment

Hive Data warehouse

HBase Data storage for distributed large tables for random, real-time read/write access to Big Data

Scribe Log collection

ZooKeeper Workflow management service

Avro Data serialization

Chukwa Data collection system to monitor distributed systems

Sqoop Open source data integration tool to integrate data into HDFS from sources outside of Hadoop

Pig High-Level Query Language Platform for analyzing huge data sets, which involves complex rules

Cassandra Highly scalable NoSQL database which brings fully distributed design & BigTable’s data model

16 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 NoSQL
• NOT relational algebra (RDBMS tables/relationships)
• SQL is not the data manipulation language

Bigtable

17 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Hybrid Systems
Business
Intelligence/
Data
Visualization

…
Enterprise
Analytics
Data
Appliance
Warehouse

Data
Integration

Data
Sources

18 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Science
Making sense of data is both Art and Science
 Generate better insights
 Gain confidence in decisions
 Visualize the data
 Understand the data and communicate that
 Learn how to learn and adapt with agility
Methods
 Data mining
 Machine learning
 Artificial intelligence
 Information retrieval
 Statistical analysis
 Gap analysis

19 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Analytics Programming Languages
Top 10
• R
• Python
• SQL
• SAS
• Java
• MATLAB
• High-level data mining suite
• UNIX shell/awk/sed
• C/C++
• Pig Latin, Hive and other Hadoop-based languages

20 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Analytics and Data Visualization

BigQuery

21 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Visualization
Context

Start Date

End Date

Competitors
Competitor1
Competitor2
Competitor3
Competitor4
Competitor5

22 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Visualization
Capacity Management Capacity Management
(Interstate Traffic – Core) (Intrastate Traffic – Transport)

= Content Provider Network Sites = Content Provider Network Sites

= Interstate Traffic (CORE) = Intrastate Traffic (Transport)
= Interconnections (ENNI, NNI, Colo, etc.) – Leased = Interconnections (ENNI, NNI, Colo,
Facilities etc.) – Leased Facilities
= Interstate Traffic (CORE)

23 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Visualization

24 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Visualization

25 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Visualization

26 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Data Visualization

27 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Ecosystem Architecture – Think Layers
Transaction Data

Big Data External Recipients

Online Transaction
Processing (OLTP) Integration Big Data • Data Hubs and Aggregators
• Compliance and eDiscovery
Mediated Processing
Online Analytical Querying
• Selective query
Processing (OLAP) & • Petabyte sorts
DW Appliances Portals • Checksum
across nodes
EDW Data • Navigational
Warehouses search
• Text mining
Interaction Data Operational Big Data Analytics
Data Stores
• Regulatory and risk
Social Media Data compliance management
Federated
Database • Innovative Risk Management-
Systems based business Models
• Community-building
Other Interaction Data Workflow • Predictive revenue and risk
Management modeling for longer time
• Clickstream Systems Big Data periods
• Text Storage • Micro- to Nano-level customer
• Images Peer-to-Peer segmentation for better
• Audio, Video • Query Availability
Integration financial service targeting
• Mobile, CDR, GPS • Fault Tolerance
• Load Distribution • Market basket analysis
• Machine, Device Personal • Social analytic solutions
• Scientific, Sensors Data • Coherent
• RFID Integration Execution

28 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Technology Adoption Maturity
Innovate
High

Big Data Analytics-driven

Business Innovation
Optimize
Business strategy modeling
Pattern analysis
and discovery Deep data mining
Business Insight and Value

Comprehensive data Predictive modeling

Implement platform for Data Scientists and decision insight
Business Analytics on
Reports, Analysis
Structured data enabled
from Big Data
Big Data store Social and sentiment Text parsing
established analysis enabled and analysis Machine Learning
For pattern discovery
First set of Business
Data integrated Unstructured
Business Technology Vendor data integrated
Learn POCs, Pilots ecosystem
Big Data Roadmap,
Technology engagement model
evaluation Big Data focus group
Technical POCs
Big Data Cluster established Big Data Architecture,
Low

Basic infrastructure design standards

Knowledge and skills

Low Big Data Technology Adoption Level High

29 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Forrester Wave – Enterprise Hadoop

30 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Gartner Magic Quadrant – DW + Hadoop

31 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Gartner Magic Quadrant – BI + Hadoop

32 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Opportunities and Implications
Business value delivery
• Research process automation, analysis and discovery
• Intellectual property
• Global, always-on data availability
• Reduced cost and rework
• Collaborative leadership
• Learning opportunities
• Rewards and fairness (vertical and horizontal equity)

33 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Clients
• Enterprise data management, architecture, governance and validation
• Financial data management
• Operational data management
• Investment data validation
• Business development analytics and decision-making
• Data query-results systems with fast processing power
• 360-degree customer views
• Social media, people-interest matching and innovation discussions

34 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Trends and Future
• Innovate
• Adapt
• Streamline
• Automate
• Consolidate
• Discover
• Extend
• Renew

35 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project Management
Where does Project Management fit in?
• Hadoop was used first in 2004 and became better known by 2007.
It is 6 years old.
• Originally, Hadoop was thought to not need project management
because it was so easy to get it together, up and running.
• Today, some clients process Petabytes of data per day globally.
Transactions reach Terabytes per second.
• The definition of Big Data has expanded to include more
technologies.
• Large infrastructure providers offer complex, expensive systems for
vertical clients.
• Implementations can involve 50-100 engineers.

36 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project Management
Big Data Analytics Project Managers are required.
• Project Managers with technical expertise and business sense are
needed.
• Expert judgment in working with the technologies is needed.
• Program Managers must help scale and synergize the efforts across
projects.
Experts are hard to find in the job market.
• Hadoop developers and administrators – 1 expert for every 2 open
positions
• Data scientists – 1 expert for every 4 open positions
• Hadoop Project and Program Managers – 1 expert for every 4 open
positions

37 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project Management
Various sources report that 65-100% of Big Data Analytics
projects fail.
• Incomplete
• Out of time
• Over budget
Sources reporting less failure than 65% include POCs, POTs, POVs and
pilots in this estimate.

38 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project Management
What happened?
• Could not hire required staff early enough or at all
• Did not anticipate current data needs and new data handling
• Did not identify sufficient components and interfaces during design
• Did not scale some of the application components to the build phase
• Did not discover critical path needs across multiple teams
• Did not load balance and test performance during development, CI and ST
• Did not have clear data architecture and governance to validate data
• Took too long to approve QA test accounts for credit cards, mobiles, CSP
traffic
• Network virtualization requirements were not clear
• Did not have time to document for support and enhancement work
• Did not have sufficient sponsorship and buy-in from client team or
stakeholders
• Budget was pulled before results could be proven
• Phase 1 infrastructure was not purchased or installed in the datacenter
• Analytics could not be implemented

39 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Program, Portfolio & PMO Management
External Impacts
• Big Data Analytics and Market Hype
• Lots of high-level information available
• Need for expertise
• Stakeholder Caution
• Need to prove methodologies, prioritization criteria and results/ROI
• Need to position for growth
• Growth because we get it
• Growth against competition’s growth
• Growth against internal resource competition and politics
• Balance centralization and decentralization of data standards and
access
• Vendor Selection
• Many attempting players with limited experience, expertise and staff
• New product lines that pair hardware and cloud software
• Marketing-oriented, low-substance presentations to clients
• Expansion of stakeholders from IT groups to Business and Marketing
groups

40 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Program, Portfolio & PMO Management
Internal Impacts
• Lack of predictive analytics about fast-moving Big Data Analytics
projects
• Lack of clarity about project synergies of expertise and reusable
components
• Lack of knowledge in selecting methodologies and coaching PMs

41 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Program, Portfolio & PMO Management
Structuring the PMO Services and Portfolios, Programs and
Projects
• Managing Requests
• Questions and consultations
• Help with clients, conferences and capabilities
• Use cases, POCs, POTs, POVs and pilots
• Staffing, training and coaching
• Technology, strategy and business value
• SLAs with clients and internal business units
• Managing the project hierarchy
• Managing methodologies
• Managing documentation

42 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Program, Portfolio & PMO Management
Organizational Adaptation – Balance of leadership and
environment
• Population Ecology – Survival of the fittest, with or without
leadership
• Life Cycles – Each cycle needs Project Management
• Creativity and entrepreneurship
• Collectivity and sophistication
• Formalization, control and efficiency
• Elaboration of structure to decentralization and expansion
• Strategic Choice – Information, locations and moments of choice
produce the actions of Prospectors, Analyzers, Defenders and
Reactors
• Symbolic Action – Defined social construction of reality and roles

43 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Program, Portfolio & PMO Management
Organizational Leadership
• Organizational Transformation
• Coherence with general theories
• Alignment with organizational frameworks and understanding
of the work
• Types of processes and ways of performing the work
• Reflective Leadership and constant communication
• Leadership as a collective, interactive practice
• Leadership in different contexts by different people
• Taking the perspective of the other people
• Leadership is a relationship
• Not a transaction
• Not a relational context such as a hierarchy or functional
role
• Transformational Leadership
• Organizational Learning
• Ambidextrous Organizations
44 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations
 Program, Portfolio & PMO Management
Thriving Projects and Project Managers
• Understand and help the client
• Collaborate with the client, elicit needs and elaborate innovations
• Deliver a rewarding scope of work and business value
• Enrich the team with good relationships and interesting, rewarding
work
• Remember work-life balance for everyone

45 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 PMBOK and Big Data Analytics
Impacted areas that require Big Data Analytics expertise
• Less delimitation of Project, Program and Portfolio Management
• Project Lifecycle
• Stakeholders and Organizational Influences
• Organizational Process Assets
• Project Management Processes
• Project Management Wisdom
Do we need a special methodology, BDA-PM?
• We need specific expertise and general understanding among
stakeholders.
• We need formalized best practices.

46 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Typical Project Lifecycle
The lifecycle can be bumpy.
• Highly iterative
• Highly exploratory
• Team is involved longer due to the need for more testing and
support
This can lead to the impression of:
• Lack of awareness and control
• Lack of communication
• Lack of planning
Set client, internal stakeholder and project team expectations
for flexibility.

47 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Typical Project Lifecycle
Initiate, Plan, Execute and Close in short phases for a series of
smaller cycles
• Use models, POCs and pilots to minimize risk and explore options
easily
• Work in parallel, overlap phases and exploit efficiencies of scale
• Build for re-use and redeployment
• Code bundling and coding maturity
• Environments (pilot, QA, staging, pre-production, production)
• Design for testability and define testing needs early to avoid
delays of a few weeks to months depending on the internal and
external services needed
• Design for easy application and infrastructure monitoring
Governance is needed. Escalations may not solve problems.
• Don’t run so fast that only critically-escalated work is completed
• Collaborate to solve problems but don’t design by negotiation
• Know when the team needs facilitation or simply sleep

48 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Organizational Process Assets
Needed OPAs do not exist in a complete, clear form
• Many traditional EDW and BI assumptions are no longer valid
• Templates may be inapplicable or need adaptation
Estimations of system volumes, capacity and load balancing
require expertise because Big Data Analytics systems process
data differently
New technology brings new knowledge
• Knowledge transfer
• Training and coaching
• Documentation of design, implementation and support
• Knowledge base development
• Coding standards
• Best practices
• Code libraries
• Reusable frameworks, tools and scripts

49 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Organizational Process Assets
Privacy and security constraints
• Delimitation of tasks, access and visibility
• More coding for data selection and masking
• Encrypted databases
• Security testing of internal code and externally facing systems
• Permission requests with formal approvals to view assets
• In-code documentation prohibited

50 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project Management Processes
Types of Processes
• CMMI is very helpful!
• Processes must be lightweight and highly facilitative
• Traditional process weight may be too heavy for Big Data Analytics
iterations
Big Data Analytics must be built as a complete ecosystem in
order to function well
• Use the roadmap, project plan and high-level WBS to identify
missing pieces
• Manage with milestones and progress toward them

Estimates have some value but may be off considerably

• Waiting for estimates does not tend to make them more accurate
• If abstraction is needed, use modified Fibonacci for Agile software
projects

51 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Schedules
Scope flexibility may be limited because a complete
application must be built
• Regulatory compliance must be in place
• Application extensions are important, including Search and other
services
• Automated testing must be coded and used
• Performance testing is required
• Monitoring and network automation are required
• Backup and failover testing is required
• One security incident can sink the product
• Users must like the product
Schedule flexibility may be limited
• Infrastructure acquisition, shipping, installation and verification
• External testing services and other new vendor contracts

52 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Quality
Quality is easier, but expertise is required to identify
requirements
• Data validation
• Data reconciliation and deduplication
• All-or-nothing transactions
• Data privacy
• ACID and BASE
• Dedicated nodes

53 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Staffing
Staffing Shortages
• Company must be willing to pay more for qualified people
• Delimit tasks on an expertise basis and push lower-level tasks to
learners
• Build people to have more expertise
• Hire people with transferable skills and train them for more

54 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Communication
Communication must be collaborative and interactive
• Define interfaces
• Do not over-ask for estimates and status updates
• Do not assume that escalation occurred appropriately
• Problem-solving may follow a path of rapid tribal knowledge
• Expertise produces efficiencies
• Most problems are real and need solutions

55 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Risk Management
Risk Management and Mitigation
• Substantial expertise in Big Data Analytics is required
• Risks can be estimated and mitigated
• Be more specific than the traditional top 5 risks

56 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Common Misconceptions
• “Big” does not mean that it’s a monolith project. Start
small and grow. Capacity management planning is easier.
• Waterfall is not required simply because it’s “Big”. Your
client who loves waterfall can still see regular reporting
without your development teams suffering through the
methodology, risks and other implications.
• Few clients are actually “Dinosaurs”. Really.
• Hype does not mean that more burn-in is needed. Hadoop
has been in use in enterprises since 2005, picked up steam
in 2007 and expanded exponentially each year since then.
• Big Data processing or storage alone does not bring the
business value.
• Analytics without data architecture, governance, validation
and quality do not bring the business value.
57 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations
 Overcoming Obstacles
DESIRED STATE
Typical obstacles that inhibit ‘Next Best Action’ or Real-Time Triggers Analytically-driven Real-
other multi-channel marketing capabilities range Submitted time offer engine
from organizational to analytical Next Best
Action

CHANGE MANAGEMENT
Approach • Organizational discipline to embed
Incremental steps reduce gaps
analytically-driven process into unified
• Subset of channels, programs and framework for offer management
members ANALYTICAL
• Process design + learning system
• Lack of channel preference segmentation,
• Framework for scaling
propensity scores by channel and program
• Absence of learning system for refinement
DATA INTEGRATION
• Data resides in silos
• Lack of 360-degree view of customer
CURRENT STATE
ORGANIZATIONAL
Data
No Action
Submitted • Programs “owned” by different groups and/or
third parties
• Different priorities and varying approaches

58 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Does Big Data Analytics fit your client?
Level of Analytics
• Client is Crawling (Storage only)
• Client is Walking (Basic Analytics for reporting what happened)
• Client is Running (Predictive Analytics for forecasting and
proaction)
Data Management
• Client has no formal data governance
• Client has centralized or decentralized data management

59 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Does Big Data Analytics fit your client?
Environment
• Big Data Analytics is only part of the client’s environment
• Client wants the cloud
• Client requires on-premise technology
• Client uses only SaaS providers and has no EDW or Data Science

Readiness for Big Data Analytics

• Client has no metrics for assessing business value
• Client has no transformational change management best practices
• Client has no in-house technical or other expertise

60 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Choosing the Best Methodology
Agile
• Scrum works best because of the iterative discovery process
• RAD
• Customer-Driven
• Stanford Advanced Project Management
• Others
Waterfall and Hybrids
• Test-integrated
• Requirements in advance

61 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Scrum and Scrum Again
Teams
• Use several scrum teams
• Development
• Unit and integration testing
• QA, automation and performance testing
• Infrastructure buildout
• Application scripting
• Application configuration of workflows, dictionary, etc.
• Network virtualization and monitoring
• Support
• Leverage knowledge as it is built

62 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Scrum and Scrum Again
Coordination
• Have a daily Scrum of Scrums and report results to management
• For 24x7 global teams, have a handoff scrum
• For the same team running double shifts, scrum morning and
evening
• For infrastructure buildout, scrum morning and evening
• Audit completions daily
• Do not ask for additional updates!

63 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Evaluating Caveats and Risks
Start with a small environment
• Do a Proof of Concept, Proof of Value or Pilot
• Use cloud services (AWS, Cloudera, etc.)
• Save money by building out the Pilot to Pre-Production and
Production
Start with a well-defined core dataset
• What data is available?
• What analytics questions are the most interesting?
• How will we add new data types?
• How will we add new analytics?

64 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Evaluating Caveats and Risks
Define well, but you do not need to know everything in
advance
• Processes and workflows
• External integration
• Component decomposition and messaging
• Necessary services
Build by layer
• Big Data takes in VVCV data
• Data processes quickly
• Analytics queries output meaningful insights
• Data is stored

65 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project: Big Data Business Case
4 weeks – Business Case Development

Use Case Business Case Business Case

Readiness Review* Use Case Selection
Definition Development Review

Customer Use Case Use Case Update Business Business Case

Preparation Matrix Definitions Case Template Reviewed with Go/No-Go
Business Sponsor Decision

Prerequisites & Use Case wireframe Business Case

Readiness Review Industry-Specific Value Accelerators Wireframe and
Questionnaire UC Repositories Guidelines

* Tasks outlined in red may be an investment and not client-billable.

66 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project: Big Data Proof of Concept
2+ weeks – Clear Scope of Use Cases and Complexity

Use Case Value

Lab Set-up Implementation
Execution Demonstration

Go Set Up POC Build Use Case Demonstrate Go/No-Go Develop and

Decision on Environment Application Business Value Decision Deploy
Business
Case

Big Data Labs Value Accelerators Value Accelerators Value Accelerators

67 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project: Big Data Strategy Assessment
Project Activities/Week 0 1 2 3 4 5 6 7 8 9 10 11 12

Introductory Workshop

Engagement Planning and Kick Off

Project Management

Big Data Strategy Assessment

Business Strategy and Priorities Discovery

Business Information Needs Assessment

Business Use Case Development

Conceptual Architecture, Vision for the Cloud,

Technology Guidance

Big Data CoE Definition and Strategy

Data Governance and Strategy

Transformational Change Management, Training

and Communication Plan

Roadmap and Implementation Plan

68 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Project: Big Data POC + Implementation
POC Analysis/Rollout Planning Implementation
Phase

~ 4 Weeks ~ 4-8 Weeks ~ 2 -12 Weeks

 Scope definition/validation  Stakeholder workshops  Rollout solution to all entities
 Data type selection  Define business objectives based on POC  Benchmark each entity
results
 Obtain sample data  Prepare markets (People)
 Detailed review of current state processes
 Analyze and correlate data  Prepare business (Process)
 Detailed review of current state technology
 Render real-time data flow  Prepare infrastructure (Technology)
(Mediation, Data storage, Analytics, Network
 Segment data interfaces, Data interfaces)  Communications plan
Activities

 Provide a Rollout baseline and plan  Detailed review of current state reporting  Align roadmap with relevant initiatives
metrics/KPIs
 Develop high level roadmap (draft)  Identify key projects and prioritize by
 Detailed gap analysis and mitigation plan ROI
 Develop high level business plan
 Define future state roadmap and tactical  Develop baseline rollout plan
 Gap analysis of current and desired
plan
analytics and business value  Develop and present results
 Develop draft rollout plan
Objective:  Identify next steps for additional data
Develop one next generation  Obtain sample data from each source types, BUs and entities
Deliverables

analytics and visualization view  Project expected results for each query type

 POC Results  Current State Maturity Mapping  Entity-by-entity rollout

 Roadmap  Gap analysis and results  Business plan benchmarking against
 Draft High-Level Business Case  Future state roadmap and tactical plan expected results
 Draft rollout plan  Executive results summary

69 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Unique Considerations
Data governance
• Data architecture
• Degree of centralization and decentralization
• New BU use case evaluation
• Metadata
• Deduplication
• Reconciliation
• Transitory data and backup
• Disaster recovery
• Site-to-site failover

70 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Unique Considerations
Security and privacy
• Encryption
• Data masking
• Server locations
• Data retention
• Customer data requests

71 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Unique Considerations
Services worth having
• Email validation
• IP validation
• GPS (latitude and longitude)
• IP location lookup
• Audit history
• Externalized metadata

72 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Unique Considerations
Application process monitoring
Network monitoring and automation
Test automation
• Regression testing
• Required manual testing for credit cards, etc.
• Required manual testing for mobiles, locality, etc.
Compliance
Metrics and KPIs

73 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Growth and Sustainability
How can systems adapt, streamline and grow?
• New data types and workflows
• Cluster management and node forecasting
• Automated deployment, network and application process
monitoring
• Replay, backup, recovery and failover
• Cloud
• Social media and mobile integration
• SDN (software discovers needs and provisions the system)

74 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Growth and Sustainability
How can we frame our thinking to optimize business value?
• What are we trying to accomplish? How is this valuable?
• What data questions do we have now? What questions may
follow?
• Can the opportunity or problem benefit from Big Data Analytics?
• Does the solution have growing room?

75 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Hands-On Exercise
• Create a repeatable model for a client’s Big Data Analytics
in a focus area that inspires you and your team:
• Vertical-specific system
• Operations and support systems
• Networking and infrastructure
• Applications and integration
• Research projects
• Investments
• Regulatory compliance
• Adaptive language and distance communications
• People and social life
• Marketing and advertising
• Education, employment and recruitment
• Continue this exercise and discussion via social media and
online applications.

76 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 References
Search online using the terms and logos in this presentation.

Big Data Analytics Project Management book by Tiffani Crawford –

http://www.amazon.com and search by title and author

Hadoop – http://www.hadoopilluminated.com/

NoSQL – http://nosql-database.org/

AWS Ecosystem – http://aws.amazon.com/

Data Science – http://www.datasciencecentral.com/

Agile – http://agilemanifesto.org/

Scrum – http://www.scrumalliance.org/

77 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations

 Speaker and Contact Information
Tiffani Crawford, PhD
408-829-7096
zapquick@aol.com
Big Data Analytics Project Management book by Tiffani Crawford – See Amazon Books at
http://www.amazon.com and search by title and author.

Tiffani Crawford, PhD, builds global Big Data Analytics systems. She has 20 years of high
technology experience with Fortune 500 companies, including Cisco Systems, Cognizant,
Bank of America, VISA/Inovant, BAE Systems, Applied Competitive Technologies, Ditech
Networks/Nuance, Big 4 financial firms, defense contractors and startups. She has worked in
seminal technology development in Big Data, analytics, cloud, networking,
telecommunications, software development, distributed multi-tier applications,
multimedia/digital, geographic information systems, intelligent transport systems, finance,
security, policy systems and structural equation modeling. She is a credited software
developer and published author. She earned her PhD from the University of Southern
California in 2005. She has also earned her Master's, Bachelor's and various technology
certifications. She is a member of PMI with various philanthropic contributions.

78 | Big Data Analytics Project Management: Methodologies, Caveats and Considerations