Transcript Advanced Analytics

Advanced Analytics
Turin
April, 2016
Index
■
Advanced Analytics Approach
– Architecture Overview
– Methodology
– Professional Skills
■
Impacted Areas / Goals
2
Advanced Analytics Approach
Business
Business Intelligence
Structured & Traditional Analysis
Answers the Questions
•
•
•
•
Asks questions
What happened?
When?
Who?
How many?
IT
Advanced Analytics Approach
(Big Data)
Discovery & Predictive Analysis
Delivers a
platform to collect
data and enable
discovery
activities
Answers the Questions
•
•
•
•
Why did it happen?
Will it happen again?
What will happen if we change a variable?
What else does the data tell us that we never thought to
ask?
IT
Structures
data/reporting to
answer business
questions
IT & Business
Explores what
questions could
be asked
Advanced Analytics approach requests:
Platform
Methodology
Skills
3
Platform
Analytic
Layer
NoSQL
Layer
HAWQ
Data Lake
Ingestion
Internal Data Sources
External Data Sources
Big data
Suite on
DCA
DIA
PHD
+
HAWQ
PHD
(Admin)
4
Methodology
Iterative Approach
Perform each phase in an
agile manner an iterate as
required
Phase 1
Problem Formulation:
Make sure to formulate a
problem that is relevant to
the goals and plain points of
the stakeholders
Phase 4
Actions
Tips and sharing of action /
strategic intervention areas,
priorities and quick-win.
Identification and sharing of
the evolutionary roadmap
about touch points or
support systems to address
business requirements.
Building a Narrative
Create a fact-based narrative
that clearly communicates
insights to stakeholders
Business Opportunity
Prioritize more relevant use
cases to cover the business
needs
Phase 2
Data and Modeling Step:
Identification of the
appropriate analysis
models based on data
acquisition processes
analysis and relevant data
sources
Phase 3
Presentation
Results representation
through dashboard and
relative report production
(positioning matrix and
SWOT)
Creativity
Take the opportunity to
innovate at every phase
5
Professional Skills – Data Scientist
Programming
 Parallelized
algorithms
Expert Statisticians
 Machine learning
Data
Scientist
 Database
practitioners
Domain Knowledge
 Process Experience
6
Impacted Areas / Goals
Quality
Manufacturing
After Sales
Connected Car
Connected Car
Supply Chain
Supply Chain
Plant
Customer
Customer
Plant
 Production processes
 Quality gates
 Order change analysis




 Eco:Drive
 MyCar
 Production Plan
 Product Engineering
 Supply Chain management
Technical service support
Warranty claim
Parts sales order
Connected Car
How to: quality production process,
recall campaign and technical
support knowledge analysis
Early Warning System
Connected Vehicles
Complexity &
Forecasting
 Goal:
−
Plant:
 Proactive prevention
Anticipate detection of emerging issues from the plant through correlations analysis between Quality
manufacturing issues and Warranty claims
−
After Sales:
 Early detection
Improve capabilities and timing of issue detection leveraging all aftersales available data
7
Impacted Areas / Goals
Quality
Manufacturing
Connected Car
After Sales
Connected Car
Supply Chain
Supply Chain
Plant
Customer
Customer
Plant
 Production processes
 Quality gates
 Order change analysis




 Eco:Drive
 MyCar
 Production Plan
 Product Engineering
 Supply Chain management
Technical service support
Warranty claim
Parts sales order
Connected Car
How to: car usage, alarm cockpit
and drive style analysis
Early Warning System
Connected Vehicles
Complexity &
Forecasting
 Goal:
−
Driver segmentation
 Improve target campaign management based on real usage of the car and better knowledge of our
customer
−
Vehicle pedigree:
 Propose maintenance services
−
Alarm warning detection:
 Improve loyalty of customer through the action based on risk score definition of failure
8
Impacted Areas / Goals
Quality
Manufacturing
Connected Car
After Sales
Supply Chain
Marketing
Supply Chain
Plant
Customer
Customer
Plant
 Production processes
 Quality gates
 Order change analysis




 Corporate Website
 Lead Generation
 Production Plan
 Product Engineering
 Supply Chain management
Technical service support
Warranty claim
Parts sales order
Connected Car
How to: best sellers configurations,
sell seasonality and processes
analysis
Early Warning System
Connected Vehicles
Complexity &
Forecasting
 Goal:
−
Product Complexity:
 Complexity reduction in order to enhance production and supply chain processes
 Support Product Manager in Product grid definition and updates
 Guide customer within decisional process while configuring the product suggesting possible/best optional of
interest
−
Order Forecasting:
 Automatic creation of Production MIX forecast with high level of accuracy
 Reduction of real orders requirements due to planning reworks
 Reduction of real material costs due to urgent transportation and scrapping costs of build out processes
9
Backup
Approaches
Hindsight
Insight
Foresight
Dynamic and
automatic plan
and actions
adjustments
based on future
events
Value of
Analytics ($)
Units sold due
to specific
marketing
campaign
proposed to
dealers
3.000 units
of 500 X
sold in the
last month
Proposed
campaign to
increase units
sold by 2%
Keys
Data Science
Business
Intelligence
How can we
make it happen?
Prescriptive
Analytics
What will
happen?
Predictive
Analytics
Why did it
happen?
Diagnostic
Analytics
What
happened?
Data
Science
Descriptive
Analytics
What is
Advanced
Analytics?
Complexity
Business Intelligence
11
Main Benefits

Analysis timing reduction and manipulations evolution

New Data Availability time-lag reduction (no preliminary transformation to load
data)

Flexible and seamless access and manipulation of structured/ unstructured data

High reduction of processing time of mining operations and application of
statistical functions

Improvement statistical performance of the models (i.e. the statistical iterations
can be multiplied for better performance)
12