SlideShare a Scribd company logo
1 of 49
Download to read offline
What Every Executive
Must Know Before Using AI
© Mobilewalla, Inc. 2018
Anindya Datta, Ph.D.
Founder, CEO, Chairman - Mobilewalla
AI is Pervasive
Investments
$26-39B
SPENT BY COMPANIES IN AI
IN 2016
$15B
FUNDING TO AI STARTUPS AS
OF 2017
AUTO, TELCO,
FINANCE
LEADING AI ADOPTERS
Trends Tokyo Olympics 2020
TOP TOPIC IN 2017
LINKEDIN INSIGHTS
200K+ ACADEMIC
PAPERS
9X INCREASE SINCE 1996
ML, DL and NLP
3 MOST IN-DEMAND JOB SKILLS
FACIAL RECOGNITION
DRIVERLESS CARS
SPECTATOR GUIDE
SYSTEM
SOURCES:McKinsey'sStateOfMachineLearningAndAI2017,CBInsightsAITrendsToWatchin
2018,LinkedInContentInsightsAnnual2017,AIIndex2017Report
© Mobilewalla, Inc. 2018
AI & Machine Learning
Strictly speaking, Machine Learning (ML) is a
sub-field of Artificial Intelligence (AI)
However, when applied to solving business
problems of interest to organizations, ML and AI
are virtually synonymous
© Mobilewalla, Inc. 2018
How
Machine
Learning
Works
© Mobilewalla, Inc. 2018
How Machine Learning Works
It all starts with a
problem
© Mobilewalla, Inc. 2018
How Machine Learning Works
Email
Automated
Spam
Filter
Spam
Non-Spam
© Mobilewalla, Inc. 2018
Two Core Components: Data & ML Algorithm
Training Data
© Mobilewalla, Inc. 2018
ML Algorithm
Binary Classification
Two Core Components: Data & ML Algorithm © Mobilewalla, Inc. 2018
Training Data ML Algorithm
Features
1. Originating Country
2. Number of Words
3. Presence of Key-phrases
TRAIN
How Machine Learning Works: #1 Train
Training Data ML Technique
© Mobilewalla, Inc. 2018
Features
1. Originating Country
2. Number of Words
3. Presence of Key-phrases
1
Binary
Classification
How Machine Learning Works: #2 Build Model
Model
Training Data ML Technique
© Mobilewalla, Inc. 2018
Intermediate Decision
Model
TRAIN
1
BUILD
MODEL
2
Features
1. Originating country
2. Number of Words
3. Presence of Key-phrases
Binary
Classification
How Machine Learning Works: #3 Validate © Mobilewalla, Inc. 2018
Features
1. Originating country
2. Number of Words
3. Presence of Key-phrases
TRAIN
1
BUILD
MODEL
2
Training Data ML Technique Intermediate Decision Model
VALIDATE
3
Hold-Out Validation Set
Binary
Classification
VALIDATE
3
How Machine Learning Works
Training Data ML Technique
© Mobilewalla, Inc. 2018
Intermediate Decision Model
Features
1. Originating country
2. Number of Words
3. Presence of Key-phrases
Prediction
Accuracy
Acceptable?
Final Decision Model
YES
NO
TRAIN
1
BUILD
MODEL
2
REFINE
4
Hold-Out Validation Set
Binary
Classification
Component 1: Training Data © Mobilewalla, Inc. 2018
ML AlgorithmTraining Data
Component 1: Training Data © Mobilewalla, Inc. 2018
Breadth
Depth
Training Data
Component 1: Training Data © Mobilewalla, Inc. 2018
Length of History
How many accurately classified
emails are included in the training
set?
What is the length of history over
which we have selected emails to
include in the training set?
Depth
Training Data
Volume
Component 1: Training Data © Mobilewalla, Inc. 2018
Length of History
How many accurately classified
emails are included in the training
set?
What is the length of history over
which we have selected emails to
include in the training set?
Depth
Training Data
Volume
Breadth Attribute Set
What specific attributes
regarding emails are part of the
training set?
Component 1: Training Data © Mobilewalla, Inc. 2018
Length of History
How many accurately classified
emails are included in the training
set?
What is the length of history over
which we have selected emails to
include in the training set?
Depth
Training Data
Volume
Breadth Attribute Set
What specific attributes
regarding emails are part of the
training set?
Features
1. Originating Country
2. Number of Words
3. Presence of Key-phrases
Component 2: ML Algorithm © Mobilewalla, Inc. 2018
ML AlgorithmTraining Data
Component 2: ML Algorithm © Mobilewalla, Inc. 2018
Linear Regression
Logistic Regression
Naive Bayes
K-Nearest Neighbours
Support Vector machines (linear kernel)
Random Forest
Gradient Boosting Tree
Reinforcement learning
ML Algorithm
Two Core Components: Data and ML Algorithm © Mobilewalla, Inc. 2018
ML AlgorithmTraining Data
Depth
Breadth
(Features)
The
question
is…
© Mobilewalla, Inc. 2018
?
What is more contributive
towards outcome?
What is more contributive
towards outcome?
What is more contributive
towards outcome?
The current AI battles
are completely fought
over techniques
Sophisticated
techniques lead to
better outcomes – this
is the common wisdom ARTIFICIAL
INTELLIGENCE
Technology Landscape
The current AI battles
are completely fought
over techniques
Sophisticated
techniques lead to
better outcomes – this
is the common wisdom ARTIFICIAL
INTELLIGENCE
Technology Landscape
AUTONOMOUS
SYSTEMS
MACHINE
LEARNING
DEEP LEARNING
NEURAL NETWORKS
PATTERN
RECOGNITION
NATURAL
LANGUAGE PROCESSING
CHAT BOTS
REAL TIME EMOTION ANALYTICS
VIRTUAL COMPANIONS
REAL TIME UNIVERSAL
TRANSLATION
THOUGHT
CONTROLLED
GAMING
NEXT GEN
CLOUD
ROBOTICS
AUTONOMOUS
SURGICAL
ROBOTICS
ROBOTIC
PERSONAL
ASSISTANTS
COGNITIVE CYBER
SECURITY
NEUROMORPHIC
COMPUTING
The current AI battles
are completely fought
over techniques
Sophisticated
techniques lead to
better outcomes – this
is the common wisdom ARTIFICIAL
INTELLIGENCE
Technology Landscape
AUTONOMOUS
SYSTEMS
MACHINE
LEARNING
DEEP LEARNING
NEURAL NETWORKS
PATTERN
RECOGNITION
NATURAL
LANGUAGE PROCESSING
CHAT BOTS
REAL TIME EMOTION ANALYTICS
VIRTUAL COMPANIONS
REAL TIME UNIVERSAL
TRANSLATION
THOUGHT
CONTROLLED
GAMING
NEXT GEN
CLOUD
ROBOTICS
AUTONOMOUS
SURGICAL
ROBOTICS
ROBOTIC
PERSONAL
ASSISTANTS
COGNITIVE CYBER
SECURITY
NEUROMORPHIC
COMPUTING
Nothing could
be further
from the
truth
Real World Use Cases
© Mobilewalla, Inc. 2018
Identifying and Building Portraits of
High-Value Customers (HVC) © Mobilewalla, Inc. 2018
Most businesses get 80% of their revenues from 20% of their customers.
These are their High-Value Customers that they want to retain, acquire and grow.
RETAIN
I must retain my
existing HVC.
ACQUIRE
I wish to acquire new
customers who are likely
to be HVC.
GROW
How do I increase the
ARPU of my HVC?
80%
20%
High-Value
CUSTOMERS
80%
20%
REVENUE
© Mobilewalla, Inc. 2018
Use Case # 1: HVC for Food Delivery Company
Training Input Features
Algorithm HVC
Portrait
Benefits
© Mobilewalla, Inc. 2018
Use Case # 1: HVC for Food Delivery Company
Training Input
Customer ID Cuisine Type Date Order Amount
1 Tandoori 08 /15/2018 $53.80
2 Pizza 08/12/2018 $27.86
3 Chinese O7/23/2018 $29.50
Customer ID Demographic
Attributes
Home
Location
Work Location Restaurant
Visit
Frequency
Supermarket
Visit
Frequency
1 Male, Single,
30-45 years
old
Paya Lebar,
Singapore
Raffles City 3x per week 1x per week
2 Male, Single,
18-34 years
old
Jurong East,
Singapore
Anson Road 4x per week 1x per week
3 Female,
Married, 25-
35 years old
Balestier,
Singapore
Marina Bay 2x per week 2x per week
© Mobilewalla, Inc. 2018
Training Input Features
Algorithm HVC
Portrait
Benefits
Data Data
Use Case # 1: HVC for Food Delivery Company
Demographic Attributes Both
Home Location Mobilewalla
Work Location Mobilewalla
Restaurant Visit Frequency Mobilewalla
Supermarket Visit Frequency Mobilewalla
Order Amt. Distribution DeliverX
Order Freq. Distribution DeliverX
© Mobilewalla, Inc. 2018
Training Input Features
Algorithm HVC
Portrait
Benefits
Data Data
Use Case # 1: HVC for Food Delivery Company
Demographic Attributes Both
Home Location Mobilewalla
Work Location Mobilewalla
Restaurant Visit Frequency Mobilewalla
Supermarket Visit Frequency Mobilewalla
Order Amt. Distribution DeliverX
Order Freq. Distribution DeliverX
• Logistic Regression (Simple)
• Gradient Boosting (Sophisticated)
© Mobilewalla, Inc. 2018
Training Input Features
Algorithm HVC
Portrait
Benefits
Data Data
Use Case # 1: HVC for Food Delivery Company
Demographic Attributes Both
Home Location Mobilewalla
Work Location Mobilewalla
Restaurant Visit Frequency Mobilewalla
Supermarket Visit Frequency Mobilewalla
Order Amt. Distribution DeliverX
Order Freq. Distribution DeliverX
• Logistic Regression (Simple)
• Gradient Boosting (Sophisticated)
• Married 25-34
• Has kids
• Working Spouse
• Home to Work Commute Distance > 15
KM
© Mobilewalla, Inc. 2018
Training Input Features
Algorithm HVC
Portrait
Benefits
Data Data
Use Case # 1: HVC for Food Delivery Company
Demographic Attributes Both
Home Location Mobilewalla
Work Location Mobilewalla
Restaurant Visit Frequency Mobilewalla
Supermarket Visit Frequency Mobilewalla
Order Amt. Distribution DeliverX
Order Freq. Distribution DeliverX
• Logistic Regression (Simple)
• Gradient Boosting (Sophisticated)
• Married 25-34
• Has kids
• Working Spouse
• Home to Work Commute Distance > 15
KM
• Increased Avg. Rev. per customer
• Increased Avg. Order Amt. per customer
• Increased Avg. LTV
• Increase in Return on Acquisition spend
© Mobilewalla, Inc. 2018
Quality of Outcome: Data (Depth) vs. Algorithm
PredictionQuality
Data Depth (Length of History)
1 Quarter 2 Quarters 3 Quarters 4 Quarters
0
1
0.7
2X
1.3X
1.1X 1.05X
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
© Mobilewalla, Inc. 2018
Quality of Outcome: Data (Depth) vs. Algorithm
PredictionQuality
Data Depth (Length of History)
1 Quarter 2 Quarters 3 Quarters 4 Quarters
0
1
0.7
2X
1.3X
1.1X 1.05X
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
✓ The deeper the data incorporated in training, the less sense it
makes to pay a premium for the technique
✓ If you haven’t incorporated deep data in training, you are likely
not getting best prediction possible
© Mobilewalla, Inc. 2018
Quality of Outcome : Data (Depth) vs. Algorithm
PredictionQuality
1 Quarter
0
1
1.4X
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
3 Quarters
© Mobilewalla, Inc. 2018
Quality of Outcome : Data (Depth) vs. Algorithm
PredictionQuality
1 Quarter
0
1
1.4X
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
3 Quarters
✓Data depth beats Algorithm hands down
© Mobilewalla, Inc. 2018
Quality of Outcome : Data (Breadth) vs. Algorithm
PredictionQuality
1
Only with DeliverX Data
With all Data
Full Featured
Wide Data
Feature Constrained
Narrow Data
1.5X
0
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
Demographic Attributes Both
Home Location Mobilewalla
Work Location Mobilewalla
Restaurant Visit Frequency Mobilewalla
Supermarket Visit Frequency Mobilewalla
Order Amt. Distribution DeliverX
Order Freq. Distribution DeliverX
© Mobilewalla, Inc. 2018
Quality of Outcome : Data (Breadth) vs. Algorithm
PredictionQuality
1
Only with DeliverX Data
With all Data
Full Featured
Wide Data
Feature Constrained
Narrow Data
1.5X
0
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
✓ Lack of breadth is very costly. Using only internal data for the project, the sophisticated technique is 50%
less effective than the simplest solution.
✓ As companies have limited breadth of 1st party customer data, enriching data sets are a must to drive
high ROI of your AI investment
© Mobilewalla, Inc. 2018
Depth vs. Breadth
PredictionQuality
1
2 Quarters of History
+
Full Features 4 Quarters of History
+
Only DeliverX Data
1.5X
0
4 Quarters of History
+
Only DeliverX Data
2 Quarters of History
+
Full Features
1.6X
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)
© Mobilewalla, Inc. 2018
Depth vs. Breadth
PredictionQuality
1
2 Quarters of History
+
Full Features 4 Quarters of History
+
Only DeliverX Data
1.5X
0
4 Quarters of History
+
Only DeliverX Data
2 Quarters of History
+
Full Features
1.6X
Logistic Regression (Simple)
Gradient Boosting (Sophisticated)✓Breadth beats depth
© Mobilewalla, Inc. 2018Use Case # 2: Brand Propensity Modelling
Training Input
Features
Algorithm Benefits
Data
• Historical QSR Visitation Data
• Brand POI data – location of the
QSR outlets
• Collaborative filtering (Simple) - CF
• Matrix Factorization (Sophisticated) –
ALS-WR
• Precise campaign planning &
targeting
• ROI maximization
• Provides guidance regarding how
consumer characteristics, both
demographic (age, gender,
marital status etc.) and
personality-specific (e.g.,
introvert/extrovert) impact
brand preference.
Results: CF and ALS WR performed equally with deeper historical data
The Results - Recall
Comparison between a simple and a sophisticated algorithm
0%
10%
20%
30%
40%
50%
60%
70%
3 Quarters -CF 1 Quarter -ALS-WR
1.3X
0%
10%
20%
30%
40%
50%
60%
70%
CF ALS-WR CF ALS-WR CF ALS-WR
1 Quarter 1 Quarter 2 Quarters 2 Quarters 3 Quarters 3 Quarters
1.42X
1.03X
1.2X
© Mobilewalla, Inc. 2018Use Case # 3: Gender Prediction
• Activity based features (1)
Device engagement, app usage, active days,
device age
• Environment based features (2)
Device type, internet connectivity, device
mobility
• App engagement based features (3)
Category of interest, demographic affinity
• Logistic Regression (Simple)
• Random forest (Sophisticated)
• Precise campaign planning &
targeting
• Personalized Messaging
• ROI maximization
Results: Logistic Regression and Random forest performed equally with
deeper historical data (3 quarters)
The Results – Nielsen Digital Ad Ratings
Comparison between a simple and a sophisticated algorithm
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Logistic
Regression
Random forest Logistic
Regression
Random forest Logistic
Regression
Random forest
1 Quarter 1 Quarter 2 Quarter 2 Quarter 3 Quarters 3 Quarters
1.22X
1.12X
1.01X
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Logistic
Regression
Logistic
Regression
Random forest Random forest
1 Quarters of
History + Full
features (1,2 & 3)
3 Quarters of
History + features
(1)
1 Quarters of
History + Full
features (1,2 & 3)
3 Quarters of
History + features
(1)
1.3X
1.22X
Breadth is
important than
history
Training Input
Algorithm Benefits
Data Features
Parting
Thoughts
© Mobilewalla, Inc. 2018
Parting Thoughts © Mobilewalla, Inc. 2018
If you are using AI/ML, are you paying attention to how your techniques
are trained?
Are you aware of the depth and breadth of first-party customer data in
your organization?
Because organizations have limited visibility into their customers, our experience is that most
first-party data is significantly breadth-constrained, and therefore depth-challenged as well.
Layering 3rd party data on your first-party information can result in meaningful increases in
quality of training data (depth and breadth), and can yield greatly increased return on your
investments in AI/ML.
48
Consumer Intelligence Solutions
SCALE
3+ years
of consumer
behavior data
1.5 B+
unique devices
30+
countries
Data from 75K
Mobile Apps
IDENTITY
Persistent Key
Online Behaviors
Offline Behaviors
Cross-Channel
INSIGHTS
SOLUTIONS
Data Enrichment
Analytics
Audience Segments
Identity
Demographic and Behavioral
Geographic and Location
Competitive
TV Attribution
© Mobilewalla, Inc. 2018

More Related Content

What's hot

Achieving GxP compliance with SAP S/4HANA in the AWS Cloud
Achieving GxP compliance with SAP S/4HANA in the AWS CloudAchieving GxP compliance with SAP S/4HANA in the AWS Cloud
Achieving GxP compliance with SAP S/4HANA in the AWS CloudCapgemini
 
Go on the offensive with c paa s
Go on the offensive with c paa sGo on the offensive with c paa s
Go on the offensive with c paa shptoga
 
APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...
APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...
APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...apidays
 
Vertical APIs as Core Product
Vertical APIs as Core ProductVertical APIs as Core Product
Vertical APIs as Core ProductZak Schwarzman
 
Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...
Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...
Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...Amazon Web Services
 
Cover story (1)- Automation
Cover story (1)- AutomationCover story (1)- Automation
Cover story (1)- Automationsmita vasudevan
 
[Minds Lab] company introduction(2020)_en
[Minds Lab] company introduction(2020)_en[Minds Lab] company introduction(2020)_en
[Minds Lab] company introduction(2020)_enSharonJung3
 
Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...
Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...
Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...accenture
 
Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...
Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...
Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...Perficient, Inc.
 
Ndimension_CorporateProfile
Ndimension_CorporateProfileNdimension_CorporateProfile
Ndimension_CorporateProfileSubhas Ghosal
 
Welcome to the API Economy: Developing Your API Strategy
Welcome to the API Economy: Developing Your API StrategyWelcome to the API Economy: Developing Your API Strategy
Welcome to the API Economy: Developing Your API StrategyMuleSoft
 
Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...
Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...
Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...Ali Saghaeian
 
Mobile Value Added Services in India
Mobile Value Added Services in IndiaMobile Value Added Services in India
Mobile Value Added Services in Indiapranavsachdeva
 
State of the SMB Market | For IP Communications and Cloud Services
State of the SMB Market | For IP Communications and Cloud ServicesState of the SMB Market | For IP Communications and Cloud Services
State of the SMB Market | For IP Communications and Cloud ServicesMetaswitch
 
I-Bytes Telecommunication &media Industry
I-Bytes Telecommunication &media IndustryI-Bytes Telecommunication &media Industry
I-Bytes Telecommunication &media IndustryEGBG Services
 
API Economy - Cuomo
API Economy - Cuomo API Economy - Cuomo
API Economy - Cuomo Prolifics
 
Communication Service Providers (CSP) and the Telecom API Ecosystem
 Communication Service Providers (CSP) and the Telecom API Ecosystem Communication Service Providers (CSP) and the Telecom API Ecosystem
Communication Service Providers (CSP) and the Telecom API EcosystemAlan Quayle
 

What's hot (20)

Achieving GxP compliance with SAP S/4HANA in the AWS Cloud
Achieving GxP compliance with SAP S/4HANA in the AWS CloudAchieving GxP compliance with SAP S/4HANA in the AWS Cloud
Achieving GxP compliance with SAP S/4HANA in the AWS Cloud
 
Go on the offensive with c paa s
Go on the offensive with c paa sGo on the offensive with c paa s
Go on the offensive with c paa s
 
APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...
APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...
APIdays Singapore 2019 - What Effective Innovative Businesses Do Differently,...
 
Vertical APIs as Core Product
Vertical APIs as Core ProductVertical APIs as Core Product
Vertical APIs as Core Product
 
Razorpay Case
Razorpay CaseRazorpay Case
Razorpay Case
 
Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...
Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...
Drive Down OpEx and Drive Up Net Promoter Score with Contextual Care (Telecom...
 
Cover story (1)- Automation
Cover story (1)- AutomationCover story (1)- Automation
Cover story (1)- Automation
 
[Minds Lab] company introduction(2020)_en
[Minds Lab] company introduction(2020)_en[Minds Lab] company introduction(2020)_en
[Minds Lab] company introduction(2020)_en
 
Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...
Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...
Cross Border: The Disruptive Frontier (Accenture Post and Parcel Industry Res...
 
Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...
Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...
Consumer Engagement with Florida Blue and Exceptional Digital Experiences Foc...
 
Ndimension_CorporateProfile
Ndimension_CorporateProfileNdimension_CorporateProfile
Ndimension_CorporateProfile
 
Welcome to the API Economy: Developing Your API Strategy
Welcome to the API Economy: Developing Your API StrategyWelcome to the API Economy: Developing Your API Strategy
Welcome to the API Economy: Developing Your API Strategy
 
Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...
Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...
Exploring Mobile Apps Categories and Successful Mobile VAS and Multimedia App...
 
Mobile Value Added Services in India
Mobile Value Added Services in IndiaMobile Value Added Services in India
Mobile Value Added Services in India
 
State of the SMB Market | For IP Communications and Cloud Services
State of the SMB Market | For IP Communications and Cloud ServicesState of the SMB Market | For IP Communications and Cloud Services
State of the SMB Market | For IP Communications and Cloud Services
 
I-Bytes Telecommunication &media Industry
I-Bytes Telecommunication &media IndustryI-Bytes Telecommunication &media Industry
I-Bytes Telecommunication &media Industry
 
API Economy - Cuomo
API Economy - Cuomo API Economy - Cuomo
API Economy - Cuomo
 
Communication Service Providers (CSP) and the Telecom API Ecosystem
 Communication Service Providers (CSP) and the Telecom API Ecosystem Communication Service Providers (CSP) and the Telecom API Ecosystem
Communication Service Providers (CSP) and the Telecom API Ecosystem
 
Mobility Solutions by Kellton Tech
Mobility Solutions by Kellton TechMobility Solutions by Kellton Tech
Mobility Solutions by Kellton Tech
 
E-Logisitics PPT
E-Logisitics PPTE-Logisitics PPT
E-Logisitics PPT
 

Similar to What Every Executive Should Know Before Implementing AI

How Deloitte Uses AI to Simplify Reporting and Increase Value
How Deloitte Uses AI to Simplify Reporting and Increase ValueHow Deloitte Uses AI to Simplify Reporting and Increase Value
How Deloitte Uses AI to Simplify Reporting and Increase ValueAmazon Web Services
 
How to Wrangle Data for Machine Learning on AWS
 How to Wrangle Data for Machine Learning on AWS How to Wrangle Data for Machine Learning on AWS
How to Wrangle Data for Machine Learning on AWSAmazon Web Services
 
Thabo Ndlela- Leveraging AI for enhanced Customer Service and Experience
Thabo Ndlela- Leveraging AI for enhanced Customer Service and ExperienceThabo Ndlela- Leveraging AI for enhanced Customer Service and Experience
Thabo Ndlela- Leveraging AI for enhanced Customer Service and Experienceitnewsafrica
 
AI and IoT innovation - an industry focus
AI and IoT innovation - an industry focusAI and IoT innovation - an industry focus
AI and IoT innovation - an industry focusAmazon Web Services
 
Deliver New Customer Experiences Through AI-enabled Chatbots
 Deliver New Customer Experiences Through AI-enabled Chatbots Deliver New Customer Experiences Through AI-enabled Chatbots
Deliver New Customer Experiences Through AI-enabled ChatbotsAmazon Web Services
 
Internet of Things e Machine Learning: i principali casi d'uso
Internet of Things e Machine Learning: i principali casi d'usoInternet of Things e Machine Learning: i principali casi d'uso
Internet of Things e Machine Learning: i principali casi d'usoAmazon Web Services
 
Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...
Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...
Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...Amazon Web Services
 
The Digital Transformation Catalyst
The Digital Transformation CatalystThe Digital Transformation Catalyst
The Digital Transformation CatalystRyan Bateman
 
Ai in retail key themes aug 2017 final
Ai in retail   key themes aug 2017 finalAi in retail   key themes aug 2017 final
Ai in retail key themes aug 2017 finalCharlotte Brook
 
Top Digital Transformation Trends (2020)
Top Digital Transformation Trends (2020)Top Digital Transformation Trends (2020)
Top Digital Transformation Trends (2020)Cygnet Infotech
 
AI in retail - Key themes October 2017
AI in retail - Key themes October 2017AI in retail - Key themes October 2017
AI in retail - Key themes October 2017Charlotte Brook
 
Ensuring Effective Service Management in the Application Economy
Ensuring Effective Service Management in the Application EconomyEnsuring Effective Service Management in the Application Economy
Ensuring Effective Service Management in the Application EconomyCA Technologies
 
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018Boaz Ziniman
 
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018Amazon Web Services
 
Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018
Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018
Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018accenture
 
Flytxt corporate brochure
Flytxt corporate brochureFlytxt corporate brochure
Flytxt corporate brochureFlytxt
 
Tectura Microchannel CRM
Tectura Microchannel CRMTectura Microchannel CRM
Tectura Microchannel CRMAtul Nebhani
 
AWS Webinar Series - Innovating the Customer Experience with Cloud and AI
AWS Webinar Series - Innovating the Customer Experience with Cloud and AIAWS Webinar Series - Innovating the Customer Experience with Cloud and AI
AWS Webinar Series - Innovating the Customer Experience with Cloud and AIAmazon Web Services
 
Enterprise Cloud Adoption
Enterprise Cloud Adoption Enterprise Cloud Adoption
Enterprise Cloud Adoption Tom Laszewski
 

Similar to What Every Executive Should Know Before Implementing AI (20)

How Deloitte Uses AI to Simplify Reporting and Increase Value
How Deloitte Uses AI to Simplify Reporting and Increase ValueHow Deloitte Uses AI to Simplify Reporting and Increase Value
How Deloitte Uses AI to Simplify Reporting and Increase Value
 
How to Wrangle Data for Machine Learning on AWS
 How to Wrangle Data for Machine Learning on AWS How to Wrangle Data for Machine Learning on AWS
How to Wrangle Data for Machine Learning on AWS
 
Thabo Ndlela- Leveraging AI for enhanced Customer Service and Experience
Thabo Ndlela- Leveraging AI for enhanced Customer Service and ExperienceThabo Ndlela- Leveraging AI for enhanced Customer Service and Experience
Thabo Ndlela- Leveraging AI for enhanced Customer Service and Experience
 
AI and IoT innovation - an industry focus
AI and IoT innovation - an industry focusAI and IoT innovation - an industry focus
AI and IoT innovation - an industry focus
 
Deliver New Customer Experiences Through AI-enabled Chatbots
 Deliver New Customer Experiences Through AI-enabled Chatbots Deliver New Customer Experiences Through AI-enabled Chatbots
Deliver New Customer Experiences Through AI-enabled Chatbots
 
Internet of Things e Machine Learning: i principali casi d'uso
Internet of Things e Machine Learning: i principali casi d'usoInternet of Things e Machine Learning: i principali casi d'uso
Internet of Things e Machine Learning: i principali casi d'uso
 
New Tools for a New World
New Tools for a New WorldNew Tools for a New World
New Tools for a New World
 
Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...
Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...
Implementation of Amazon Connect, Powered by Accenture (FSV306-S) - AWS re:In...
 
The Digital Transformation Catalyst
The Digital Transformation CatalystThe Digital Transformation Catalyst
The Digital Transformation Catalyst
 
Ai in retail key themes aug 2017 final
Ai in retail   key themes aug 2017 finalAi in retail   key themes aug 2017 final
Ai in retail key themes aug 2017 final
 
Top Digital Transformation Trends (2020)
Top Digital Transformation Trends (2020)Top Digital Transformation Trends (2020)
Top Digital Transformation Trends (2020)
 
AI in retail - Key themes October 2017
AI in retail - Key themes October 2017AI in retail - Key themes October 2017
AI in retail - Key themes October 2017
 
Ensuring Effective Service Management in the Application Economy
Ensuring Effective Service Management in the Application EconomyEnsuring Effective Service Management in the Application Economy
Ensuring Effective Service Management in the Application Economy
 
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
 
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
Starting your Cloud Transformation Journey - Tel Aviv Summit 2018
 
Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018
Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018
Disrupting the Disruptors: Accenture Post and Parcel Industry Research 2018
 
Flytxt corporate brochure
Flytxt corporate brochureFlytxt corporate brochure
Flytxt corporate brochure
 
Tectura Microchannel CRM
Tectura Microchannel CRMTectura Microchannel CRM
Tectura Microchannel CRM
 
AWS Webinar Series - Innovating the Customer Experience with Cloud and AI
AWS Webinar Series - Innovating the Customer Experience with Cloud and AIAWS Webinar Series - Innovating the Customer Experience with Cloud and AI
AWS Webinar Series - Innovating the Customer Experience with Cloud and AI
 
Enterprise Cloud Adoption
Enterprise Cloud Adoption Enterprise Cloud Adoption
Enterprise Cloud Adoption
 

Recently uploaded

Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationRidwan Fadjar
 
Bluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfBluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfngoud9212
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraDeakin University
 
Pigging Solutions in Pet Food Manufacturing
Pigging Solutions in Pet Food ManufacturingPigging Solutions in Pet Food Manufacturing
Pigging Solutions in Pet Food ManufacturingPigging Solutions
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):comworks
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsRizwan Syed
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 3652toLead Limited
 
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024BookNet Canada
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticscarlostorres15106
 
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Wonjun Hwang
 
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks..."LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...Fwdays
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxnull - The Open Security Community
 
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphSIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphNeo4j
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersThousandEyes
 
Pigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping ElbowsPigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping ElbowsPigging Solutions
 
Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024Scott Keck-Warren
 

Recently uploaded (20)

Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
Transcript: New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
My Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 PresentationMy Hashitalk Indonesia April 2024 Presentation
My Hashitalk Indonesia April 2024 Presentation
 
Bluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdfBluetooth Controlled Car with Arduino.pdf
Bluetooth Controlled Car with Arduino.pdf
 
Artificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning eraArtificial intelligence in the post-deep learning era
Artificial intelligence in the post-deep learning era
 
Pigging Solutions in Pet Food Manufacturing
Pigging Solutions in Pet Food ManufacturingPigging Solutions in Pet Food Manufacturing
Pigging Solutions in Pet Food Manufacturing
 
CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):CloudStudio User manual (basic edition):
CloudStudio User manual (basic edition):
 
Scanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL CertsScanning the Internet for External Cloud Exposures via SSL Certs
Scanning the Internet for External Cloud Exposures via SSL Certs
 
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
Tech-Forward - Achieving Business Readiness For Copilot in Microsoft 365
 
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
New from BookNet Canada for 2024: BNC BiblioShare - Tech Forum 2024
 
Vulnerability_Management_GRC_by Sohang Sengupta.pptx
Vulnerability_Management_GRC_by Sohang Sengupta.pptxVulnerability_Management_GRC_by Sohang Sengupta.pptx
Vulnerability_Management_GRC_by Sohang Sengupta.pptx
 
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmaticsKotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
Kotlin Multiplatform & Compose Multiplatform - Starter kit for pragmatics
 
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
Bun (KitWorks Team Study 노별마루 발표 2024.4.22)
 
DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
Hot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort Service
Hot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort ServiceHot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort Service
Hot Sexy call girls in Panjabi Bagh 🔝 9953056974 🔝 Delhi escort Service
 
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks..."LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
 
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptxMaking_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
Making_way_through_DLL_hollowing_inspite_of_CFG_by_Debjeet Banerjee.pptx
 
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge GraphSIEMENS: RAPUNZEL – A Tale About Knowledge Graph
SIEMENS: RAPUNZEL – A Tale About Knowledge Graph
 
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for PartnersEnhancing Worker Digital Experience: A Hands-on Workshop for Partners
Enhancing Worker Digital Experience: A Hands-on Workshop for Partners
 
Pigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping ElbowsPigging Solutions Piggable Sweeping Elbows
Pigging Solutions Piggable Sweeping Elbows
 
Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024Advanced Test Driven-Development @ php[tek] 2024
Advanced Test Driven-Development @ php[tek] 2024
 

What Every Executive Should Know Before Implementing AI

  • 1. What Every Executive Must Know Before Using AI © Mobilewalla, Inc. 2018 Anindya Datta, Ph.D. Founder, CEO, Chairman - Mobilewalla
  • 2. AI is Pervasive Investments $26-39B SPENT BY COMPANIES IN AI IN 2016 $15B FUNDING TO AI STARTUPS AS OF 2017 AUTO, TELCO, FINANCE LEADING AI ADOPTERS Trends Tokyo Olympics 2020 TOP TOPIC IN 2017 LINKEDIN INSIGHTS 200K+ ACADEMIC PAPERS 9X INCREASE SINCE 1996 ML, DL and NLP 3 MOST IN-DEMAND JOB SKILLS FACIAL RECOGNITION DRIVERLESS CARS SPECTATOR GUIDE SYSTEM SOURCES:McKinsey'sStateOfMachineLearningAndAI2017,CBInsightsAITrendsToWatchin 2018,LinkedInContentInsightsAnnual2017,AIIndex2017Report © Mobilewalla, Inc. 2018
  • 3. AI & Machine Learning Strictly speaking, Machine Learning (ML) is a sub-field of Artificial Intelligence (AI) However, when applied to solving business problems of interest to organizations, ML and AI are virtually synonymous © Mobilewalla, Inc. 2018
  • 5. How Machine Learning Works It all starts with a problem © Mobilewalla, Inc. 2018
  • 6. How Machine Learning Works Email Automated Spam Filter Spam Non-Spam © Mobilewalla, Inc. 2018
  • 7. Two Core Components: Data & ML Algorithm Training Data © Mobilewalla, Inc. 2018 ML Algorithm
  • 8. Binary Classification Two Core Components: Data & ML Algorithm © Mobilewalla, Inc. 2018 Training Data ML Algorithm Features 1. Originating Country 2. Number of Words 3. Presence of Key-phrases
  • 9. TRAIN How Machine Learning Works: #1 Train Training Data ML Technique © Mobilewalla, Inc. 2018 Features 1. Originating Country 2. Number of Words 3. Presence of Key-phrases 1 Binary Classification
  • 10. How Machine Learning Works: #2 Build Model Model Training Data ML Technique © Mobilewalla, Inc. 2018 Intermediate Decision Model TRAIN 1 BUILD MODEL 2 Features 1. Originating country 2. Number of Words 3. Presence of Key-phrases Binary Classification
  • 11. How Machine Learning Works: #3 Validate © Mobilewalla, Inc. 2018 Features 1. Originating country 2. Number of Words 3. Presence of Key-phrases TRAIN 1 BUILD MODEL 2 Training Data ML Technique Intermediate Decision Model VALIDATE 3 Hold-Out Validation Set Binary Classification
  • 12. VALIDATE 3 How Machine Learning Works Training Data ML Technique © Mobilewalla, Inc. 2018 Intermediate Decision Model Features 1. Originating country 2. Number of Words 3. Presence of Key-phrases Prediction Accuracy Acceptable? Final Decision Model YES NO TRAIN 1 BUILD MODEL 2 REFINE 4 Hold-Out Validation Set Binary Classification
  • 13. Component 1: Training Data © Mobilewalla, Inc. 2018 ML AlgorithmTraining Data
  • 14. Component 1: Training Data © Mobilewalla, Inc. 2018 Breadth Depth Training Data
  • 15. Component 1: Training Data © Mobilewalla, Inc. 2018 Length of History How many accurately classified emails are included in the training set? What is the length of history over which we have selected emails to include in the training set? Depth Training Data Volume
  • 16. Component 1: Training Data © Mobilewalla, Inc. 2018 Length of History How many accurately classified emails are included in the training set? What is the length of history over which we have selected emails to include in the training set? Depth Training Data Volume Breadth Attribute Set What specific attributes regarding emails are part of the training set?
  • 17. Component 1: Training Data © Mobilewalla, Inc. 2018 Length of History How many accurately classified emails are included in the training set? What is the length of history over which we have selected emails to include in the training set? Depth Training Data Volume Breadth Attribute Set What specific attributes regarding emails are part of the training set? Features 1. Originating Country 2. Number of Words 3. Presence of Key-phrases
  • 18. Component 2: ML Algorithm © Mobilewalla, Inc. 2018 ML AlgorithmTraining Data
  • 19. Component 2: ML Algorithm © Mobilewalla, Inc. 2018 Linear Regression Logistic Regression Naive Bayes K-Nearest Neighbours Support Vector machines (linear kernel) Random Forest Gradient Boosting Tree Reinforcement learning ML Algorithm
  • 20. Two Core Components: Data and ML Algorithm © Mobilewalla, Inc. 2018 ML AlgorithmTraining Data Depth Breadth (Features)
  • 22. What is more contributive towards outcome?
  • 23. What is more contributive towards outcome?
  • 24. What is more contributive towards outcome?
  • 25. The current AI battles are completely fought over techniques Sophisticated techniques lead to better outcomes – this is the common wisdom ARTIFICIAL INTELLIGENCE Technology Landscape
  • 26. The current AI battles are completely fought over techniques Sophisticated techniques lead to better outcomes – this is the common wisdom ARTIFICIAL INTELLIGENCE Technology Landscape AUTONOMOUS SYSTEMS MACHINE LEARNING DEEP LEARNING NEURAL NETWORKS PATTERN RECOGNITION NATURAL LANGUAGE PROCESSING CHAT BOTS REAL TIME EMOTION ANALYTICS VIRTUAL COMPANIONS REAL TIME UNIVERSAL TRANSLATION THOUGHT CONTROLLED GAMING NEXT GEN CLOUD ROBOTICS AUTONOMOUS SURGICAL ROBOTICS ROBOTIC PERSONAL ASSISTANTS COGNITIVE CYBER SECURITY NEUROMORPHIC COMPUTING
  • 27. The current AI battles are completely fought over techniques Sophisticated techniques lead to better outcomes – this is the common wisdom ARTIFICIAL INTELLIGENCE Technology Landscape AUTONOMOUS SYSTEMS MACHINE LEARNING DEEP LEARNING NEURAL NETWORKS PATTERN RECOGNITION NATURAL LANGUAGE PROCESSING CHAT BOTS REAL TIME EMOTION ANALYTICS VIRTUAL COMPANIONS REAL TIME UNIVERSAL TRANSLATION THOUGHT CONTROLLED GAMING NEXT GEN CLOUD ROBOTICS AUTONOMOUS SURGICAL ROBOTICS ROBOTIC PERSONAL ASSISTANTS COGNITIVE CYBER SECURITY NEUROMORPHIC COMPUTING Nothing could be further from the truth
  • 28. Real World Use Cases © Mobilewalla, Inc. 2018
  • 29. Identifying and Building Portraits of High-Value Customers (HVC) © Mobilewalla, Inc. 2018 Most businesses get 80% of their revenues from 20% of their customers. These are their High-Value Customers that they want to retain, acquire and grow. RETAIN I must retain my existing HVC. ACQUIRE I wish to acquire new customers who are likely to be HVC. GROW How do I increase the ARPU of my HVC? 80% 20% High-Value CUSTOMERS 80% 20% REVENUE
  • 30. © Mobilewalla, Inc. 2018 Use Case # 1: HVC for Food Delivery Company Training Input Features Algorithm HVC Portrait Benefits
  • 31. © Mobilewalla, Inc. 2018 Use Case # 1: HVC for Food Delivery Company Training Input Customer ID Cuisine Type Date Order Amount 1 Tandoori 08 /15/2018 $53.80 2 Pizza 08/12/2018 $27.86 3 Chinese O7/23/2018 $29.50 Customer ID Demographic Attributes Home Location Work Location Restaurant Visit Frequency Supermarket Visit Frequency 1 Male, Single, 30-45 years old Paya Lebar, Singapore Raffles City 3x per week 1x per week 2 Male, Single, 18-34 years old Jurong East, Singapore Anson Road 4x per week 1x per week 3 Female, Married, 25- 35 years old Balestier, Singapore Marina Bay 2x per week 2x per week
  • 32. © Mobilewalla, Inc. 2018 Training Input Features Algorithm HVC Portrait Benefits Data Data Use Case # 1: HVC for Food Delivery Company Demographic Attributes Both Home Location Mobilewalla Work Location Mobilewalla Restaurant Visit Frequency Mobilewalla Supermarket Visit Frequency Mobilewalla Order Amt. Distribution DeliverX Order Freq. Distribution DeliverX
  • 33. © Mobilewalla, Inc. 2018 Training Input Features Algorithm HVC Portrait Benefits Data Data Use Case # 1: HVC for Food Delivery Company Demographic Attributes Both Home Location Mobilewalla Work Location Mobilewalla Restaurant Visit Frequency Mobilewalla Supermarket Visit Frequency Mobilewalla Order Amt. Distribution DeliverX Order Freq. Distribution DeliverX • Logistic Regression (Simple) • Gradient Boosting (Sophisticated)
  • 34. © Mobilewalla, Inc. 2018 Training Input Features Algorithm HVC Portrait Benefits Data Data Use Case # 1: HVC for Food Delivery Company Demographic Attributes Both Home Location Mobilewalla Work Location Mobilewalla Restaurant Visit Frequency Mobilewalla Supermarket Visit Frequency Mobilewalla Order Amt. Distribution DeliverX Order Freq. Distribution DeliverX • Logistic Regression (Simple) • Gradient Boosting (Sophisticated) • Married 25-34 • Has kids • Working Spouse • Home to Work Commute Distance > 15 KM
  • 35. © Mobilewalla, Inc. 2018 Training Input Features Algorithm HVC Portrait Benefits Data Data Use Case # 1: HVC for Food Delivery Company Demographic Attributes Both Home Location Mobilewalla Work Location Mobilewalla Restaurant Visit Frequency Mobilewalla Supermarket Visit Frequency Mobilewalla Order Amt. Distribution DeliverX Order Freq. Distribution DeliverX • Logistic Regression (Simple) • Gradient Boosting (Sophisticated) • Married 25-34 • Has kids • Working Spouse • Home to Work Commute Distance > 15 KM • Increased Avg. Rev. per customer • Increased Avg. Order Amt. per customer • Increased Avg. LTV • Increase in Return on Acquisition spend
  • 36. © Mobilewalla, Inc. 2018 Quality of Outcome: Data (Depth) vs. Algorithm PredictionQuality Data Depth (Length of History) 1 Quarter 2 Quarters 3 Quarters 4 Quarters 0 1 0.7 2X 1.3X 1.1X 1.05X Logistic Regression (Simple) Gradient Boosting (Sophisticated)
  • 37. © Mobilewalla, Inc. 2018 Quality of Outcome: Data (Depth) vs. Algorithm PredictionQuality Data Depth (Length of History) 1 Quarter 2 Quarters 3 Quarters 4 Quarters 0 1 0.7 2X 1.3X 1.1X 1.05X Logistic Regression (Simple) Gradient Boosting (Sophisticated) ✓ The deeper the data incorporated in training, the less sense it makes to pay a premium for the technique ✓ If you haven’t incorporated deep data in training, you are likely not getting best prediction possible
  • 38. © Mobilewalla, Inc. 2018 Quality of Outcome : Data (Depth) vs. Algorithm PredictionQuality 1 Quarter 0 1 1.4X Logistic Regression (Simple) Gradient Boosting (Sophisticated) 3 Quarters
  • 39. © Mobilewalla, Inc. 2018 Quality of Outcome : Data (Depth) vs. Algorithm PredictionQuality 1 Quarter 0 1 1.4X Logistic Regression (Simple) Gradient Boosting (Sophisticated) 3 Quarters ✓Data depth beats Algorithm hands down
  • 40. © Mobilewalla, Inc. 2018 Quality of Outcome : Data (Breadth) vs. Algorithm PredictionQuality 1 Only with DeliverX Data With all Data Full Featured Wide Data Feature Constrained Narrow Data 1.5X 0 Logistic Regression (Simple) Gradient Boosting (Sophisticated) Demographic Attributes Both Home Location Mobilewalla Work Location Mobilewalla Restaurant Visit Frequency Mobilewalla Supermarket Visit Frequency Mobilewalla Order Amt. Distribution DeliverX Order Freq. Distribution DeliverX
  • 41. © Mobilewalla, Inc. 2018 Quality of Outcome : Data (Breadth) vs. Algorithm PredictionQuality 1 Only with DeliverX Data With all Data Full Featured Wide Data Feature Constrained Narrow Data 1.5X 0 Logistic Regression (Simple) Gradient Boosting (Sophisticated) ✓ Lack of breadth is very costly. Using only internal data for the project, the sophisticated technique is 50% less effective than the simplest solution. ✓ As companies have limited breadth of 1st party customer data, enriching data sets are a must to drive high ROI of your AI investment
  • 42. © Mobilewalla, Inc. 2018 Depth vs. Breadth PredictionQuality 1 2 Quarters of History + Full Features 4 Quarters of History + Only DeliverX Data 1.5X 0 4 Quarters of History + Only DeliverX Data 2 Quarters of History + Full Features 1.6X Logistic Regression (Simple) Gradient Boosting (Sophisticated)
  • 43. © Mobilewalla, Inc. 2018 Depth vs. Breadth PredictionQuality 1 2 Quarters of History + Full Features 4 Quarters of History + Only DeliverX Data 1.5X 0 4 Quarters of History + Only DeliverX Data 2 Quarters of History + Full Features 1.6X Logistic Regression (Simple) Gradient Boosting (Sophisticated)✓Breadth beats depth
  • 44. © Mobilewalla, Inc. 2018Use Case # 2: Brand Propensity Modelling Training Input Features Algorithm Benefits Data • Historical QSR Visitation Data • Brand POI data – location of the QSR outlets • Collaborative filtering (Simple) - CF • Matrix Factorization (Sophisticated) – ALS-WR • Precise campaign planning & targeting • ROI maximization • Provides guidance regarding how consumer characteristics, both demographic (age, gender, marital status etc.) and personality-specific (e.g., introvert/extrovert) impact brand preference. Results: CF and ALS WR performed equally with deeper historical data The Results - Recall Comparison between a simple and a sophisticated algorithm 0% 10% 20% 30% 40% 50% 60% 70% 3 Quarters -CF 1 Quarter -ALS-WR 1.3X 0% 10% 20% 30% 40% 50% 60% 70% CF ALS-WR CF ALS-WR CF ALS-WR 1 Quarter 1 Quarter 2 Quarters 2 Quarters 3 Quarters 3 Quarters 1.42X 1.03X 1.2X
  • 45. © Mobilewalla, Inc. 2018Use Case # 3: Gender Prediction • Activity based features (1) Device engagement, app usage, active days, device age • Environment based features (2) Device type, internet connectivity, device mobility • App engagement based features (3) Category of interest, demographic affinity • Logistic Regression (Simple) • Random forest (Sophisticated) • Precise campaign planning & targeting • Personalized Messaging • ROI maximization Results: Logistic Regression and Random forest performed equally with deeper historical data (3 quarters) The Results – Nielsen Digital Ad Ratings Comparison between a simple and a sophisticated algorithm 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Logistic Regression Random forest Logistic Regression Random forest Logistic Regression Random forest 1 Quarter 1 Quarter 2 Quarter 2 Quarter 3 Quarters 3 Quarters 1.22X 1.12X 1.01X 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Logistic Regression Logistic Regression Random forest Random forest 1 Quarters of History + Full features (1,2 & 3) 3 Quarters of History + features (1) 1 Quarters of History + Full features (1,2 & 3) 3 Quarters of History + features (1) 1.3X 1.22X Breadth is important than history Training Input Algorithm Benefits Data Features
  • 47. Parting Thoughts © Mobilewalla, Inc. 2018 If you are using AI/ML, are you paying attention to how your techniques are trained? Are you aware of the depth and breadth of first-party customer data in your organization? Because organizations have limited visibility into their customers, our experience is that most first-party data is significantly breadth-constrained, and therefore depth-challenged as well. Layering 3rd party data on your first-party information can result in meaningful increases in quality of training data (depth and breadth), and can yield greatly increased return on your investments in AI/ML.
  • 48. 48 Consumer Intelligence Solutions SCALE 3+ years of consumer behavior data 1.5 B+ unique devices 30+ countries Data from 75K Mobile Apps IDENTITY Persistent Key Online Behaviors Offline Behaviors Cross-Channel INSIGHTS SOLUTIONS Data Enrichment Analytics Audience Segments Identity Demographic and Behavioral Geographic and Location Competitive TV Attribution