The Economics of Data, Analytics and Digital Transformation
Transcript of The Economics of Data, Analytics and Digital Transformation
Source: Bill Schmarzo “Big Data MBA” Course Curriculum “Economics of Data™“ Playing Cards
The Economics of Data, Analytics and Digital Transformation
Bill SchmarzoHitachi Vantara Chief Innovation Officer
Honorary Professor, National University of Ireland-GalwayUniversity of San Francisco, Executive Fellow
Twitter: @schmarzo
Bill Schmarzo “Big Data MBA” Curriculum
Data isthe new oil...and muchmore…
Bill Schmarzo “Big Data MBA” Curriculum
BUSINESS OPTIMIZATION
BUSINESS INSIGHTSBUSINESS
MONITORING
INSIGHTSMONETIZATION
DIGITALTRANSFORMATION
PrescriptiveRecommendations
Big Data Economics
Key Business
Use Cases
How Effective is Your Organization at Leveraging
Data and Analytics to Power your Business Models?
Big Data Business Model Maturity Index
ANALYTICSCHASM
Bill Schmarzo “Big Data MBA” Curriculum
Business Initiative
Stakeholders
Decisions (Use Cases)
Analytic Assets
Data & Instrumentation
Architecture & Technology
Data Science Value Engineering Framework
© Hitachi, Ltd. 2018. All rights reserved.
Leveraging Data Science to Transform Your Economic Value Curve
5
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
Maintenance Spend
Up
tim
e %
Lo Hi
Hi
Up1
Up2
C1 C2
Economic Value Curve ChallengeEconomic Value Curve determines optimization point between multiple variables. Unfortunately, Law of Diminishing Returns dictates that additional spend yields marginal improvements.
Maintenance costs could include direct and indirect costs such as work hours, overtime costs, extra parts and inventory, extra consumables, and the costs associated with fixing parts that were not going to break
Δ
Δ
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
Transforming the Economic Value Curve
LoHi
Up2
C2
Up3
Original Economic Value Curve
New Economic Value Curve
The way to beat Law of Diminishing Returns is to leverage analytics to create new Economic Value Curve; that
is, increase Uptime (from Up2 to Up3) with less Maintenance spend (from C2 to C3).
C3
Maintenance costs could include direct and indirect costs such as work hours, overtime costs, extra parts and inventory, extra consumables, and the costs associated with fixing parts that were not going to break
Maintenance Spend
Up
tim
e %
Hi
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
How Might COVID-19 Digitally Transform Healthcare Industry?
Source: https://ourworldindata.org/the-link-between-life-expectancy-and-health-spending-us-focus
COVID-19 will force healthcare organizations to digitally transform by uncovering granular patient, disease,
treatment, wellness, doctor, hospital, etc. insights to re-engineer healthcare services
• Smart Hospitals
• Intelligent Healthcare apps
• Precision medicine
• Personalized preventative care
• Personalized welfare
• Remote wellness diagnostics
• Predictive world health
• ML-assisted imaging diagnostics
• AI-based Digital Assistants
• Prescriptive health monitoring
• Digital therapeutics
• Concierge care
US Economic Value CurveROW Economic Value Curve
Source: Bill Schmarzo “Big Data MBA” Course Curriculum “Economics of Data™“ Playing Cards
Rapid exploration, rapid testing, failure-embracing, continuously-learning and adapting “Scientific Method”
REPEAT
Step 1: Define Hypothesis (Decision)to test or Prediction to make
Step 3: Prepare data; Schema-on-query
Step 4: Visualize the data (Tableau, Pentaho, ggplot2,…)
Step 5: Build analytic models (TensorFlow, Python, Jupyter…)
Step 2: Gather data…and more data (Data Lake: SQL + NoSQL)
HistoricalGoogle Trends
PhysicianNotes
Local Events
Weather Forecast CDC
LawsonEpic
Kronos
Step 6: Evaluate model “goodness of fit” (coefficients, confidence levels)
Source: “Scientific Method: Embrace the Art of Failure”, University of San Francisco School of Management Big Data MBA
Data Science Collaborative Engagement Process
Source: Bill Schmarzo “Big Data MBA” Course Curriculum “Economics of Data™“ Playing Cards
The Art of “Thinking Like A Data Scientist”
5 IdentifyData Sources
76Map Scores to
Recommendations 8Group Metrics
Into ScoresIdentify
Recommendations
1Identify
Business Initiative
Identify Analytic Entities
32Identify
Stakeholders
4Identify / Prioritize
Use Cases
Source: Bill Schmarzo “Big Data MBA” Course Curriculum11
PROPENSITIES
TENDENCIES
INCLINATIONS
PREFERENCES
BEHAVIORS
INTERESTS
PASSIONS
ASSOCIATIONS
AFFILIATIONS
BIASES
AFFINITIES
TASTES
You Don’t Monetize Volume of Data; You Monetize Granularity
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
Entity-level Asset Models: Patient Analytic Profile
External Patient Data• Diet History (DietPlanner,
MyFitnessPal)
• Physical Exercise History (MapMyRun, FitBit)
• Mental Acuity History (Lumosity, CogniFit)
• Stress History (Stress Doctor, Happify)
• Emotional History (Text, Social)
• Vices History
• Vacation / Relaxation History
• Others…
Patient Care Data• Demographic
• Behavioral Demographics
• Psychographics
• Patient care / treatment history
• Patient vital stats history
• Physician / Nurse care notes
• Patient comments
• Pharmacy
• Others…
Schmarzo Patient Profile Score Variance Trend
Health Score 92 1.89
Wellness Score 92 1.85
Diet Score 67 3.25
Exercise Score 82 2.25
Stress Score 65 1.90
COVID19 At-Risk Score 22 2.35
Cancer At-Risk Score 14 1.74
Pulmonary At-Risk Score 02 1.15
Oncology At-Risk Score 08 1.20
Heart Attack At-Risk Score 09 1.25
Stroke At-Risk Score 06 1.10
….
CONFIDENTIAL – For use by Hitachi and Disney employees under NDA only. © Hitachi, Ltd. 2018. All rights reserved.
Exploiting the Economics of Data, Analytics and Digital Transformation
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
USF Economic Value of Data Research• Data an asset that never depletes, never wears out, and can be used
across unlimited use cases at zero marginal cost
Customer point of sales data
Sales
Promotional effectiveness
+2.5%
• Economic Multiplier Effect: ratio of the impact of an incremental
increase in investment on the resulting incremental increase in value
• Accounting: “Value in Exchange” methodology for determining asset valuation based upon the acquisition cost of an asset
• Economics: “Value in Use” methodology for determining asset valuation
Marketing
Customer acquisition
+2.0%
Call Center
Customer retention
+3.5%
Product Dev
New product intro
+2.6%
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
Detailed historical transactions coupled with internal unstructured and publicly-available
data sources
Data transformed into analytic assets (scores, rules, propensities, segments, recommendations)
ANALYTICSDATA
Clusters of decisions around common subject area in support of organization’s key business initiatives
USE CASES
Solving Technology-to-Business Linkage Challenge
Bill Schmarzo “Big Data MBA” Curriculum
To Change The Game, Change Your Frame…
If you buy a Tesla today, I believe you're buying an appreciating asset, not a depreciating asset
– Elon MuskTesla CEO
“”
Bill Schmarzo “Big Data MBA” Curriculum
Role of Artificial Intelligence Agents in Learning
Defining the Utility Function is Critical for Autonomous
“Success”
Get reward New state
StateTake actionAI Agent Environment
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
The more the product gets used… the more accurate, more robust, more predictive and consequently more
valuable the product becomes. The value of the product appreciates, not depreciates, with usage
What Powers an “Appreciating” Asset?
Deep Reinforcement Learning ModelRational AI State
Bill Schmarzo “Big Data MBA” Curriculum
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
Tesla Autopilot Continuous Learning Environment
•Millions of Miles from 500,000+ Tesla Cars• Billions of Miles from Autopilot
Simulator
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
steering
electrical
engine or motor
fuelpassenger
suspensionIOT EdgeAnalytics
Real-Time Data Streaming
Sensor Data
MachineLearning
Real-timeDecisions
Modeling(Deep
Learning)
Time-seriesData Mgmt
Defining the Utility Function is Critical for Autonomous
“Success”
Rational AI Agent
State Environment
Deep Reinforcement Learning Model
Bill Schmarzo “Big Data MBA” Curriculum
Exploit Economics of Compounding Improvements
Driving and
operational data;
“edge” use cases
Backpropagate
learnings (updated
models)
Law of 1% Compounding 1.01 ^ 365 = 37.8x
Source: Bill Schmarzo “Big Data MBA” Course Curriculum
Incremental “Use Case-by-Use Case” approach to building, reusing and refining data and analytic assets
enables Attribution of use case Financial Value to enabling Data Sources and Analytic Assets
Data Sources
Analytic Assets
Use Cases
Building and Valuing Digital Assets Use Case-by-Use Case
New
Reused
Not Used
Asset Legend
(1) Vendor Quality$60M
A B C$20M $20M $20M
A B$30M $30M
(2) Vendor Reliability$20M
A B C
D
$10M($30M)
$10M
A B$10M($40M)
C$10M
(3) Optimize Inventory$60M
A B C$15M($45M)
$15M($35M)
D E F$15M $15M
A B$20M($60M)
C$20M($50M)
D$20M
…
© Hitachi Vantara Corporation 2019. All Rights Reserved© Hitachi Vantara Corporation 2019. All Rights Reserved
“Economies of Learning” More Powerful than “Economies of Scale”
21
Source: Bill Schmarzo “Big Data MBA” Course Curriculum “Economics of Data™“ Playing CardsUse Cases
Cum
ulat
ive V
alue
($$$
)
Lo
Hi
“Economies of Learning” more powerful than the “Economies of Scale”: The more the data and
analytics get used, the more accurate, more effective, more predictive, more valuable they become
Effect #3: Economic Value Accelerates• Refining Analytic Module effectiveness ripples thru
previous use cases that use same Analytic Module –The Google TensorFlow Effect
Effect #2: Economic Value Grows• Data and analytic module re-use shrinks time-to-
value and de-risks use cases
Effect #1: Marginal Costs Flatten• Reusing “curated” data and analytic modules
reduces marginal costs for new use case (no data silos or orphaned analytics)
Schmarzo Economic Digital Asset Valuation Theorem
23Source: Bill Schmarzo “Big Data MBA” Course Curriculum
BILL SCHMARZOHitachi Vantara, Chief Innovation Officer
University San Francisco School of Management, Executive Fellow
Honorary Professor, National University of Ireland-Galway
Top-ranking Blogsv To Achieve Big Data’s Potential, Get It into the Boardroom
v Big Data Business Model Maturity Index
v 6 Laws of Digital Transformation
v History Lesson on Economic-Driven Business Transformation
v User Experience: The New King of the Business
v IOT: Transitioning from Connected to “Smart”
v Learning How to “Think Like a Data Scientist”
Contact [email protected]
Find me on Twitter: @schmarzo
Thank You!