TDT4259 –Applied Data Science

Lecture 9: Decision making with

Data Science
Nisha Dalal
Adj. Associate Professor

[email protected]
 2

Important Deadlines
When you will need to deliver or complete a task

1 20/9 Register yourself/group and the company/dataset for group assignment

2 30/10 Deliver individual assignment

3 27/11 (Noon) Deliver presentation and report for group assignment

 9

Data Analysis bringing Business Value: Target

• Targeting: New parents

• Strategy for new parents:

• Get them to buy diapers
• Personalized offers on diapers and baby products

• Indicators of new parents:

• Diapers, wipes, other baby products etc.
• Open birth records

• How to stay ahead in the game?

• Detect pregnancy?
• Detect trying to get pregnant?
The age of data
 11

The Age of Data

How data is leading the next digital transformation

Suggested Talk

Philip Evans: How data will transform business

79% of enterprise
executives agree that companies
that do not embrace Big Data will
lose their competitive position
and could face extinction
(Accenture, 2018)
 12

Types of value creation from Big Data

5 ways in which Big Data can create value

TRANSPARENCY DECISION MAKING MARKETING PERFORMANCE INNOVATION

Big data can substantially Big data allows organizations to Big data can enable Big data enables companies
Making big data more easily
improve decision making, create highly specific organizations to collect more to create new products and
accessible to relevant
minimize risks, and unearth segmentations and to tailor accurate and detailed services, enhance existing
stakeholders in a timely
valuable insights that would products and services precisely performance data (in real or ones, and invent entirely
manner
otherwise remain hidden to meet those needs near real time) new business models.

Wamba, S. F., Akter, S., Edwards, A., Chopin, G., & Gnanzou, D. (2015). How ‘big
data’can make big impact: Findings from a systematic review and a longitudinal case
study. International Journal of Production Economics, 165, 234-246.
5 Ways in which Big Data add Value
An illustrative example of cases
 14

Transparency
New forms of information presentation

Delta Airlines
Delta is one of the three major United States airlines with headquarters in
Atlanta, Georgia, USA.

With over 130 million bags checked per year, the company held a lot of
tracking data about bags and became the first major airline to allow
customers to track their bags from mobile devices in real-time by adding
RFID’s in the bag tags.

To date, the app has been downloaded over 11 million times and gives
customers much greater peace of mind while traveling while also
differentiating Delta as a customer-centric company.
 15

Decision Making
Real-time and better-informed decisions

UPS
UPS more than 18 million package and document deliveries every day and
over 4 billion items shipped per year through almost 100,000 vehicles.

On-truck telematics and advanced algorithms help with routes, engine idle
time, and predictive maintenance. It also uses ORION (On-road Integrated
Optimization and Navigation) to create optimal routes for delivery drivers
from the data supplied by customers, drivers and the vehicles.

Since starting the program, the company has saved over 148 million liters of
fuel, avoided driving 582 million kilometers and reduced engine idle time by
206 million minutes.
 16

Marketing
Getting to know customers better

Amazon
Segmentation of population to customize offerings and marketing campaigns
(consumer goods, retail, social, clinical data, etc).

The company reported a 29% sales increase to $12.83 billion during its
second fiscal quarter, up from $9.9 billion during the same time in 2012.

The retail giant’s recommendation algorithms are based on seemingly few

elements: a user’s purchase history, items in their shopping cart, items
they’ve rated and liked, and what other customers have viewed and
purchased.

The importance of suggesting the right item to the right user can be gauged
by the fact that 35% of all sales are estimated to be generated by the
recommendation engine.
 17

Performance
Fine-tuning performance

Southwest Airlines
Exploration for, and discovery of, new needs, can drive organizations to fine tune for
optimal performanceand efficiency (employee data).

Southwest Airlines, is using speech analytics to extract data-rich info from live-
recorded interactions between customers and personnel to get a better
understanding of their customers.

Speech analytics help the airline to have maximum information about the customer
beforehand.

Whenever a customer calls his voice is organized by a speech engine and his recent
queries and important data are displayed which helps in improving interaction.
 18

Innovation
Driving new products and services

BlackRock Investment Group

BlackRock, Inc. is an American global investment management corporation based in
New York City and is the world's largest asset manager with $6.29 trillion in assets
under management as of December 2017.

BlackRock has set up a new center dedicated to research in artificial intelligence,

underscoring the heightened interest among asset managers in how machine
learning can revolutionize many facets of the investment industry.

BlackRock is already analyzing traffic through corporate websites as an indicator to

the future growth of a company. It also employs text analysis tools to comb through
transcripts of earnings calls, looking for signs of whether management team is
positive or negative about their company’s growth prospects.
 19

Some other use cases

• Healthcare: In healthcare, data science is used to analyze patient data, medical records, and test
results to aid in disease diagnosis. Machine learning models can help predict diseases like diabetes or
cancer, providing early detection and better treatment options.

• Retail: Retailers use data science to optimize inventory management. By analyzing historical sales data
and current trends, they can make data-driven decisions about stock levels and replenishment,
reducing costs and preventing stockouts.

• Finance: In the financial industry, credit scoring models use data science to assess the
creditworthiness of individuals or businesses. These models analyze credit history, income, and other
factors to determine the risk of lending money.

• E-commerce: Companies like Amazon and Netflix use data science to provide personalized
recommendations to users. These systems analyze user behavior and preferences to suggest products
or content, ultimately increasing user engagement and sales.
 20

Some other use cases

• Transportation: In smart cities, data science is used to monitor and manage traffic flow. Real-time data
from cameras, sensors, and GPS devices are analyzed to make decisions about traffic light timings,
route optimization, and congestion management.

• Manufacturing: Manufacturers use data science for quality control. By analyzing sensor data and
production metrics, they can detect defects early in the manufacturing process and reduce waste and
rework.

• Marketing: Marketers use data science to segment customers into groups based on behavior,
demographics, or preferences. This segmentation helps tailor marketing campaigns and content to
specific customer groups.

• Agriculture: Farmers use data science to predict crop yields. Factors like weather data, soil conditions,
and historical crop data are analyzed to make informed decisions about planting and harvesting
schedules.
 21

Importance of explainability
• Explainability refers to the ability to understand and interpret the reasoning behind the decisions
made by machine learning models or data-driven systems.

Source
 22

Importance of explainability
• Explainability refers to the ability to understand and interpret the reasoning behind the decisions
made by machine learning models or data-driven systems.

Source
 23

Importance of explainability
• Trust and Accountability
• Compliance and Regulations
• Detecting Bias and Fairness
• Error Identification and Improvement
• User Acceptance
• Ethical Considerations
• Domain Expertise
• Continuous Improvement
• Communication

Explainability in data-driven decision making is not only about understanding how models work but also
about ensuring fairness, ethics, compliance, and user trust.
 24

Levels of explainability
• Model-Specific explainability
• Feature Importance
• Local and Global Interpretability
• Counterfactual explanations
• Ethical and Fairness explanations
• Multimodal Explanations
• Ensemble Model explainability
• Human-Centric Explanations

The choice of explainability layer depends on the specific use case, the complexity of the model, the
intended audience, and the ethical and regulatory considerations. A combination of these layers may be
necessary to provide comprehensive insights and ensure the responsible and effective use of machine
learning models.
 25

Nisha Dalal
Questions & Discussion [email protected]