How do you assess your Machine Learning capabilities and use it to make decisions on your assets? George W. Earle, Director, Global Commercial Office, ThoughtWorks helps finds answers.
4. WHAT I’LL COVER
INTRODUCTION to Problems that are becoming untenable?
SITUATION Requiring Machine Learning Capability if Possible
ACTIONS and Decisions to be Taken RE: Machine Learning
RESULTS of Asset Owners, Does ML Help?
POINTS to Take HomeKey
QUESTIONS
6. INTRODUCTION TO PROBLEMS IN TWO
INDUSTRIES
INDUSTRIAL MACHINERY
Thoughtworks helps the Machinery
industries (Heavy Equip,
Manufacturing, Utilities) with their
Aftermarket Sales & Asset
Operations challenges
AIRLINES
We also help the Airline
industry with their Asset
Utilization challenges and in
the future, Asset Availability
7. From the 1800s through the present, there has been an accelerating evolution of
interaction, networking and device technology, and the near future will bring interaction
modes that are increasingly transparent and always available.
As industries migrate to these new technologies, enterprise technology ecosystems
must incorporate them or risk losing their customer relationships to disruptive tech
startups.
8. ASSETS (EQUIPMENT, DEVICES,
CONTROLLERS) & IOT
What is being done with all this data?
1. Proverbial Heads promptly enter the sand - or
2. Dismember and Analyze it to death - or
3. Intelligently Connect and Feed-it-back
From the 1800s through the present, there has been an accelerating evolution of
interaction, networking and device technology, and the near future will bring interaction
modes that are increasingly transparent and always available.
As industries migrate to these new technologies, enterprise technology ecosystems
must incorporate them or risk losing their customer relationships to disruptive tech
startups.
10. SITUATION REQUIRING MACHINE
LEARNING CAPABILITY - 1
Equipment Aftermarket Situation 1
Client needed to provide equipment
servicing opportunities for their dealers,
based on pre-fixed schedules &
equipment usage data
Existing solution wasn't scaling, didn't
provide reliable answers
11. SITUATION REQUIRING MACHINE
LEARNING CAPABILITY - 2
Equipment Servicing Situation 2
Airlines have 100s+ planes with
unplanned maintenance events
Different Airlines have different
unplanned maintenance rates 2.5 Zettabytes / year
are not going to be BI analyzed
14. ACTIONS AND DECISIONS TO BE
TAKEN RE: MACHINE LEARNING
What is all the fuss about?
Current solutions don’t adjust, use
estimates, intuition, and numerical
analysis.
Machine Learning mathematics use
equations that through SW can self
adjust.
So Machine Learning isn’t a fad, since
it is modern mathematics.
“Use of algorithms will … [be] part of tomorrow’s
management vocabulary as, say, profit”
Ram Charan – Executive Consultant, author of the
bestselling book, Execution
15. 15
MACHINE LEARNING BASICS
Statistics have fixed parameters. Machine
Learning models have self-adjusting parameters.
Supervised Unsupervised Reinforcement
Machine
Learning
16. ACTIONS AND DECISIONS TO BE
TAKEN RE: MACHINE LEARNING
One dim regression
underestimates risk, hides insight
Phase 1.0 - 4.0 Strategy
Accumulation Stage – Big Data1.0
Prediction Stage - Insight2.0
Prescription Stage – Behaviour3.0
Machine Collaboration - Action4.0
McKinsey & Company
18. 18
RESULTS OF ASSET OWNERS, DOES ML HELP?
Equipment Aftermarket Situation 1
1 Created scalable solution
that provided reliable
predictive answers
2 Forecasted servicing time for
customers using arithmetic
progression + sales feedback
3 Success measured in dealer
take-up
19. 19
RESULTS OF ASSET OWNERS, DOES ML HELP?
Equipment Servicing Situation 2
1 Application* receives real
time engine, aircraft data,
flight plans
2 Optimizes fuel economy,
anticipates maintenance,
avoids downtime
3 Success from carrying
optimal fuel level, pre-
positioning parts ahead
*MS Azure/ Rolls Royce
21. WHAT ADVANCED ASSET
OPPORTUNITES ARE CLEAR AND
PRESENT?
You are acquiring a ton of data and
can’t get the right insights
• Focus on the 1.0 to 4.0 strategy to output
intelligent data and analytics
• Augment staff Subject Matter Expertise
with intelligent analytics
• Tie analytics for intelligent Actions directly
to dramatic improvement in financial KPIs
• Look for partners who create competitive &
personalized solutions
DATA
INSIGHTS
ACTIONS
24. George W. Earle
Director Global Commercial Office
ThoughtWorks Inc.
Gearle@thoughtworks.com
DRIVE VALUE FROM ASSETS
Hinweis der Redaktion
Let’s start with an industry where the problem is self apparent since their assets are in the public view and they still won’t learn.
Some have 80-90% of their planes generating and unplanned event
Some have only 10% of their planes…
Meanwhile… 2.5 Zettabytes / year are not going to be analyzed.
Phase 1.0 - 4.0 Strategy
1.0 – Accumulation Stage - collecting ‘Big Data’ in your organization, you may already be acting on this
2.0 – Prediction Stage - obtaining new insights from messy big data, using Machine Learning
3.0 – Prescription Stage – CSuite understanding allows companies to affect customer behavior in the future, act on machine learning
4.0 – Machine Collaboration -- Humanity, Augmented interprets machine patterns and recommends a course of action. SMEs use insights from previous stages for better outcomes. In the above example, use the insights to create new financial products for customers in need.