Machine learning in retail

CW
IN
CAPGEMINI
WEEK OF
INNOVATION
NETWORKS
Machine Learning in Retail
Shailesh Agaskar/Pradeep Koona

Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 2Presentation Title | Author | Date © 2018 Capgemini. All rights reserved.
Market Basket Analysis
Evolution of Retail using ML
Which of my products tend to be
purchased together?
What do other people like this person
tend to like/buy/watch?
Which other link does a customer
navigate to after selecting a item in his
shopping basket on a online shopping
store?

Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 3
✓ Enhanced Inventory Management
✓ Customer Segmentation
✓ Targeted Marketing
✓ Product Rating and Ranking
✓ Recommender Systems
✓ Dynamic Pricing
✓ Process Automation
✓ Payment Services and Fraud Prevention
✓ Location of new stores
✓ Automated Customer Service
✓ Churn Modeling
Applications of Machine Learning in Retail

Customer Segmentation
Recognize similar customers
and their tendency to buy
certain products.
Enhanced Inventory
Management
Recommender Systems
Market Basket Analysis
ML Applications in Retail sector
Place related items together to
encourage sales.
Identify under performing
fixtures and run promotions.
Helps identify which products
trend to be purchased together
and lift sales.

Inventory Management using Association Rules
Enhanced Inventory Management
Iteration 1
{ A } , { B } , { C } are frequent and
{ D } is not frequent.
Iteration 2
It considers only
{ A,B } , { A,C } , { B,C } for evaluation
{ A,D } , { B,D } , { C,D } are not considered.
Iteration 3
It checks if { A , B , C } is not frequent
and say { A,C } is not frequent, the Algorithm stops.
Specifically designed for mining over transactions in
databases
Apriori Algorithm
Apriori algorithm uses the components like support,
confidence and lift to identify the association rules that
describe the relationship between customer transactions.
Support: It is the percentage of an item that appears in a set
of transactions.
Confidence: It is the percentage of transactions that contain
both X and Y out of all the transactions that contain X
Lift: It is a measure of how much more likely one item is to
be purchased given another item has already been
purchased.

Groceries transactions set from a super market
for 60 days
Input
Processed transactions are fed to the Machine
learning model.
Output
Item relations are derived from the algorithm.
Enhanced Inventory Management – Use Case

01
02
03
Output Relations advantages to
instore management
Enhanced
Inventory
Management
Arrangement of Items
Related items are displayed together
▪ Encourages sales
▪ Helps customers to pick related items together
Sales and Promotions
Underperforming fixtures can be identified
▪ Discounts and freebies can be applied
▪ Sales promotions for a particular day or time can be
announced.
Stock Management
Sales forecast for a particular time period is accessed
▪ Take Smart stocking decisions
▪ Avoid procuring few items which don’t belong to the forecast

Input (Transaction set to Algorithm) Output (Relations identified)
Enhanced Inventory Management

Microsoft’s Azure MLaaS can extend
the output of Assosiation rules as a
service
Recommender systems have proved to
drive sales of majority of E-Commerce
applications by more than 35%
E-Commerce Web applications can
encourage sales using Recommender
systems
Recommender System – Azure ML

Item – to – Item Collaborative Filtering Personalized Notifications
Recommender System – Use Case
A more specific relationship between two items can be found and
this relationship uplifts confidence of another new purchase.
Related items if identified and shown at the right time encourage
sales and drive additional revenue.
Make seasonal and trend based supply decisions simpler.
Recommender System in E-Commerce - Microsoft Azure ML
Dormant users are activated by sending push
notifications.
App engagement can be achieved by sending
the push notifications at the most suitable time
for the customer.
A E-commerce application which sells bikes and biking gear online.
We identify the relationships between two specific items using Azure Machine Learning – collaborative filtering technique
based on past purchasing trends and recommend the related items to the customer before he does a check-out on the active
purchases.

E-commerce Biking gear Application

Azure ML – MLaaS service

Presentation Title | Author | Date © 2018 Capgemini. All rights reserved.
This message contains information that may be privileged or confidential and is
the property of the Capgemini Group.
Copyright © 2018 Capgemini. All rights reserved.
A global leader in consulting, technology services and digital transformation, Capgemini is
at the forefront of innovation to address the entire breadth of clients’ opportunities in the
evolving world of cloud, digital and platforms. Building on its strong 50-year heritage and
deep industry-specific expertise, Capgemini enables organizations to realize their business
ambitions through an array of services from strategy to operations. Capgemini is driven
by the conviction that the business value of technology comes from and through people.
It is a multicultural company of 200,000 team members in over 40 countries. The Group
reported 2017 global revenues of EUR 12.8 billion.
About Capgemini
Learn more about us at
www.capgemini.com

Machine learning in retail

More Related Content

What's hot (20)

Similar to Machine learning in retail (20)

More from Capgemini (20)

Recently uploaded (20)

Machine learning in retail