A Major Project Report

On

GROCERY OFFER SUGGESTION SYSTEM FOR

RETAIL MARKETS
Submitted to JNTU HYDERABAD
In Partial Fulfillment of the requirements for the Award of Degree of

BACHELOR OF TECHNOLOGY
IN
COMPUTER SCIENCE AND ENGINEERING

Submitted
By

Jahnavi Adabala (178R1A0561)

Sowmya Chikkam (178R1A0569)
Namrata Rikkibe (178R1A05A3)
Sowmya Maale (178R1A0596)

Under the Esteemed guidance of

Dr. M. Kumara Swamy
Professor, Department of CSE

Department of Computer Science & Engineering

CMR ENGINEERING COLLEGE

(Approved by AICTE, NEW DELHI, Affiliated to JNTU, Hyderabad)
Kandlakoya, Medchal Road, R.R. Dist. Hyderabad-501 401)
2020-2021
 CMR ENGINEERING COLLEGE
(Accredited by NBA, Approved by AICTE NEW DELHI, Affiliated to JNTU, Hyderabad)
Kandlakoya, Medchal Road, Hyderabad-501 401

Department of Computer Science & Engineering

CERTIFICATE

This is to certify that the project entitled “ GROCERY OFFER SUGGESTION SYSTEM FOR
RETAIL MARKETS” is a bonafide work carried out by
Jahnavi Adabala (178R1A0561)
Sowmya Chikkam (178R1A0569)
Namrata Rikkibe (178R1A05A3)
Sowmya Maale (178R1A0596)
in partial fulfillment of the requirement for the award of the degree of BACHELOR OF
TECHNOLOGY in COMPUTER SCIENCE AND ENGINEERING from CMR Engineering
College, affiliated to JNTU, Hyderabad, under our guidance and supervision.

The results presented in this project have been verified and are found to be satisfactory. The
results embodied in this project have not been submitted to any other university for the award of
any other degree or diploma.

Internal Guide Major Project Coordinator Head of the Department

Dr. M. Kumara Swamy Mrs. G. Sumalatha Dr. Sheo Kumar

Professor Associate Professor Professor & H.O.D
Department of CSE Department of CSE Department of CSE,
CMREC, Hyderabad CMREC, Hyderabad CMREC, Hyderabad
 DECLARATION

This is to certify that the work reported in the present project entitled "GROCERY OFFER
SUGGESTION SYSTEM FOR RETAIL MARKETS" is a record of bonafide work done by us in
the Department of Computer Science and Engineering, CMR Engineering College, JNTU Hyderabad.
The reports are based on the project work done entirely by us and not copied from any other source. We
submit our project for further development by any interested students who share similar interests to
improve the project in the future.

The results embodied in this project report have not been submitted to any other University or Institute
for the award of any degree or diploma to the best of our knowledge andbelief.

Jahnavi Adabala (178R1A0561)

Sowmya Chikkam (178R1A0569)
Namrata Rikkibe (178R1A05A3)
Sowmya Maale (178R1A0596)
 ACKNOWLEDGEMENT

We are extremely grateful to Dr. A. Srinivasula Reddy, Principal and Dr. Sheo Kumar, HOD,
Department of CSE, CMR Engineering College for their constant support.

I am extremely thankful to Dr. M. Kumara Swamy, Professor, Internal Guide, Department of CSE,
for her constant guidance, encouragement and moral support throughout the project.

I will be failing in duty if I do not acknowledge with grateful thanks to the authors of the
references and other literatures referred in this Project.

I express my thanks to all staff members and friends for all the help and co-ordination extended in
bringing out this Project successfully in time.

Finally, I am very much thankful to my parents who guided me for every step.

Jahnavi Adabala (178R1A0561)

Sowmya Chikkam (178R1A0569)
Namrata Rikkibe (178R1A05A3)
Sowmya Maale (178R1A0596)
 CONTENTS
TOPIC PAGENO

1. INTRODUCTION
1.1 Introduction & Objectives 1
1.2 Purpose of the Project 2
1.3 Existing system & Disadvantages 2
1.4 Proposed system with Features 3

2. LITERATURE SURVEY 6

3. SOFTWARE & HARDWARE REQUIREMENTS

3.1. Software requirements 9
3.2. Hardware requirements 37

4. CODING AND IMPLEMENTATION

4.1. Sample Code 39
4.2. Data Dictionary 71

5. OUTPUT SCREENS 72

6. CONCLUSION 73

7. FUTURE ENHANCEMENTS 74

8. BIBLIOGRAPHY AND REFERENCES 75

 ABSTRACT

Face recognition technology enables offline grocery stores to do what the online shopping
recommendation systems have been doing for years.
Using a combination of in-store cameras and facial recognition software, shops can now
easily asses the purchase data of the customers and provide offers on products based on the
purchase history.
We are designing a system where cameras at entry recognize the customer and links his
previous purchase history stored in the database.
 1. INTRODUCTION

1.1 Introduction and Objectives

All online retail apps have proved that, when a retailer knows its customers, it can serve them
more effectively. But how does that work in a physical, offline grocery retail environment? One
way is to integrate a facial recognition system. Face recognition technology enables offline
grocery stores to do what the online shopping recommendation systems have been doing for
years i.e., cameras in offline grocery shops identify their shoppers/customers, link them to past
purchases and generate personalized product recommendations based on the data.

Using a combination of in-store cameras and facial recognition software, shops can now easily
assess the purchase data of the customers and recommend products based on the history. This
core data can be collected at each stage of the customer journey, tracking shoppers (and how they
interact with a store) from entry to checkout. Not only can face recognition technology identify
and classify customers, it can help retailers optimize and plan their product offerings.

Recommendation system is a collection of tools and techniques to provide products or services

suggestions for users. The existence of this system allows companies to develop a marketing
strategy, attract more customers, and increase sales. Therefore, many companies try to
implement a recommendation system for their business interest. Recommendation system has
been applied in a variety of industries. It can be found in the entertainment domain (music,
movies, TV shows, books), news or tourism sites, ecommerce, e-library, and e-learning systems.

E-commerce has already implemented the recommendation system with many benefits such as
boosting up customer level of interaction, increasing sales, the diversity of items sold, customer
satisfaction or loyalty, and also understanding customers’ demand better. Such benefits are
expected to be achieved in traditional or offline retail stores.

The objectives of this project are:

● to learn what is an open CV.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 1

 ● to know how to do image processing.

● to know how to read a webcam using OpenCV.

● to know how to retrieve the data from Database.

● to know more about IBM Cloud Services.

● to work on Open cv face Detection algorithms.

1.2 Purpose of the Project

We are designing a system where cameras at entry recognize the customer and link his previous
purchase history stored in the database. Based on that purchase history he will get a
recommendation of the offers on the product which he wishes to buy. And also displays on the
screen at the entrance as soon as he steps into the store.

Customer details are procured when he first visits the Shop. His details like name, phone number
will be taken at the entry point and also captures the images of person Purchase data will be
taken at the exit.

1.3 Existing System and Disadvantages

In grocery stores, large-scale transaction data with identification, such as point of sales (POS)
data, is being accumulated as a result of the introduction of frequent shopper programs (FSPs).
The accumulated POS data have been used to examine customer shopping behavior, especially
by professionals in the marketing field1. Although the recommendations based on this data are
often adopted in e-commerce shopping stores, they are rarely introduced in face-to-face selling,
such as in brick-and-mortar grocery stores. Therefore, introducing a system based on these
recommendations to grocery stores could induce customers to visit the store to make a purchase.
We propose two recommended systems based on stored POS data. The first system gathers the e-
mail address during the registration procedure and directly determines recommended products
based on the stored POS data and sends reminders with discount information to customers by
email. When constructing this system for grocery stores, the sparsity of evaluation values
presents a problem. Evaluation values are constructed based on customers’ purchase frequency
of product

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 2

items and are very sparse. This is because individual customers only purchase very few of the
total number of product items a store sell. We alleviated the problem of data sparsity by
proposing a system in which recommended product items are determined by a two-step
procedure. First, the system determines recommended product categories enabling the store
manager to determine recommended product items according to the circumstances of the
particular grocery store; for example, specific product items may have to be cleared out from
inventory or could be purchased at a lower price than usual from the manufacturer. Discount
information relating to the recommended product items are sent to remind customers by email
similar to the direct system. The problem of data sparsity is often addressed technically by
adopting singular value decomposition (SVD) in the recommended system3, 4, 5. In our work
we evaluate various recommended methods, including SVD-type recommendations, based on
real POS data. A representative recommendation method is (i) user-based collaborative filtering
(CF). A representative SVD-type recommendation method is (ii) recommendation based on
evaluation values reconstructed by SVD, which factorizes the original evaluation matrix, thereby
reducing it to low rank matrices and reconstructing the evaluation matrix. We also evaluate
another type of SVD recommendation, (iii) recommendation by item-user similarity by SVD and
(iv) recommendation by using a combination of CF and SVD. In addition, we evaluate (v)
recommendation by nonlinear principal component analysis (NL-PCA) 6, also known as
sandglass-type neural networks (SNN). The results of numerical experiments show that
recommendation by SVD reconstruction is suitable for the recommendation of product items via
the direct recommendation method, whereas CF is suitable for the recommendation of product
categories via the two-step recommendation method.

1.4 Proposed System with Features

Using a combination of in-store cameras and facial recognition software, shops can now easily
assess the purchase data of the customers and recommend products based on the history. This
core data can be collected at each stage of the customer journey, tracking shoppers (and how they
interact with a store) from entry to checkout. Not only can face recognition technology identify
and classify customers, it can help retailers optimize and plan their product offerings. We are
designing a system where cameras at entry recognize the customer and link his previous
purchase history stored in the database. Based on that purchase history he will get a
recommendation of the

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 3

offers on the product which he wishes to buy. And also displays on the screen at the entrance as
soon as he steps into the store.

Customer details are procured when he first visits the Shop. His details like name, phone number
will be taken at the entry point and also captures the images of person Purchase data will be
taken at the exit.

The project flow will have the following activities:

● Customers enter the store and store cameras identify the face of the customer.

● The identified face is stored in the cloud object storage service and the respective image
URL is stored in Cloudant DB.

● Fetch the stored image.

● This image is then sent to the custom model build for face recognition using Visual
recognition.

● Visual recognition recognizes the customer.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 4

 ● Once the customer is recognized we retrieve the previous products purchased by the
customers from cloudant DB.

● Based on previous purchases recommendations are showcased on the UI displayed at the

entrance for a particular customer.

● If the customer is new then he gets registered at the entrance.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 5

 2. LITERATURE SURVEY

Kolesar and Galbraith contend that e-retail encompasses three main activities: (1) a product
search activity that provides detailed information on the products under evaluation, which is
usually referred to as a product-evaluation or information-gathering (IG) facility; (2) an online
purchase function that facilitates consumer interaction by reducing the transaction costs; and (3)
a product delivery capability that facilitates the final product’s distribution to consumers. Darley
et al. present an overall review to understand to what extent the current marketing and consumer
behavior body of literature can be transferred to the analysis of online consumer behavior and
preferences. The paper uses the model proposed by Engel, Kollat and Blackwell as the analytical
framework to synthesize the findings from the literature. The Engel–Kollat–Blackwell (EKB)
model proposes five core stages of the decision-making process, as follows: (1) problem
recognition; (2) search; (3) alternative evaluation purchase; (4) choice; and (5) outcomes. The
authors contend that the underlying alternative evaluation depends on internal and external
factors. The internal factors are categorized into three different classes: (1) cognitive or beliefs;
(2) affective or attitudes; and (3) cognitive or intentions. On the other hand, the external or
environmental factors are subdivided into four categories: (1) individual differences or
characteristics such as motives, values, lifestyle and personality; (2) socio-cultural factors, such
as culture, social class, reference groups and family; (3) situational and economic factors; and (4)
online attributes, such as a Web site’s quality, a Web site’s interface, Web site satisfaction and
Web site experience. The model is finally extended with the consequences of the decision-
making process, in which the relevance finally focuses on the satisfaction/dissatisfaction
construct. The inherent model recognizes the complexity and the multidimensionality of online
consumer behavior, and that—besides being a comprehensive approach—other interactions,
antecedents and consequences can be finally envisaged and included. In this respect, the authors
highlight that empirically tested constructs and relationships are preferred to conceptual
approaches. Two key constructs are found: (1) personal satisfaction and Sustainability; and (2)
trust, security and company reputation. These two constructs are clearly related to other three
core constructs that have gained significant attention from both scholars and practitioners:
purchase, repurchase and product return. The authors conclude that the research agenda could
benefit from exploring the differences in consumers’ preferences between brick-and-mortar and
online outlets (brick to click

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 6

consumer behavior). In this respect, online consumers tend to be very heterogeneous in different
contexts. For example, Rohm and Swaminathan [18] find four stereotypes of e-grocery grocery
shoppers: (1) convenience shoppers; (2) variety seekers; (3) balanced buyers; and (4) store-
oriented shoppers.

As several studies indicate(e.g. Deloitte & Harrison Group, 2010; POPAI, 2011), grocery
shoppers have begun to find ways to spend less and reduce risk in the midst of the current
economic and financial crisis. They have learned new tactics to save money on supermarket
purchases and manage their household pantry, while shopping trips have also become more
careful and focused (Deloitte & Harrison Group, 2010). Consumers’ grocery shopping routine
now regularly includes strategic and tactical features like clarifying wants versus needs, delaying
gratification, lowering quality requirements, frequent channel, store and brand switching, an
intense use of coupons, loyalty cards, shopping lists and other promotional offers, stockpiling
and increasing purchase of private label products, among others (Deloitte & Harrison Group,
2010; POPAI, 2011). Furthermore, consumers are becoming increasingly less loyal to national
14 brands and also less likely to engage in impulse buying or new product trials, as the new aim
for grocery shopping is household gratification while maintaining quality but minimizing
expenditure (Deloitte & Harrison Group, 2010). Consumers are no longer afraid or ashamed to
be seen shopping for a bargain, often viewing price as the single most important factor in
choosing among retail brands and also a motive to patronize multiple stores, formats and retail
brands (POPAI, 2011). Shoppers are also increasingly synergizing between the off- and on-line
channels, in order to maximize the value of their purchases (POPAI, 2011; Deloitte & Harrison
Group, 2010). To the same end, they are also becoming more receptive to new electronic
shopping tools and savvier as to which fit better their purchase needs and plans, increasingly
seeking all sorts of information resources available to gain more control over their shopping
experience (POPAI, 2011). And while these new approaches and strategies are mostly based on
cutting down expenditure, most consumers still do not feel like they are sacrificing much, and
thus show no intention of returning to old shopping habits when the economy recovers (POPAI,
2011; Deloitte & Harrison Group, 2010). According to A.C. Nielsen’s annual report on consumer
confidence (2010), Portuguese consumers are no exception to this scenario. As fellow shoppers
worldwide, they are also changing their spending habits – e.g., eating out less, buying less
garments and more private labels, being more concerned about energy and gas spending –, and

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 7

show no intention of returning to old shopping habits.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 8

Online grocery stores were amongst the first Internet start-ups to be launched in the late 1990’s.
Nevertheless, it was only very late that this type of business was able to gain enough traction
amongst North-American and European consumers, currently accounting for only a small portion
of total online retail sales in many countries (KPMG International, 2012; Lim, Windows &
Hooker, 2009; Ramus & Nielsen, 2005). Nevertheless, after books, clothing and other product
categories, grocery shopping is emerging as the next fastest growing category in online retailing
(KPMG International, 2012). As such, the strong predicted growth of the online grocery channel
in Western countries presents good business opportunities for brand manufacturers and retailers
alike (Lim, Windows & Hooker, 2009). Due to the current economic conditions, however, this
study found that more consumers from all income levels are beginning to buy groceries online,
looking to benefit from the variety of promotional offers that tend to be exclusive to this retail
format. Additionally, the KPMG International (2012) study uncovered that consumers are
gradually shifting from personal computers to mobile phones and tablets as the preferred device
for online shopping. Consequently, the continued development of mobile applications and its
increased convenience of use and associated services should help increase even more the
occurrence of online grocery shopping (KPMG International, 2012). Moreover, the referred
study defends that online grocery stores offer the contemporary, time and money tight customers
a way to conveniently search and acquire products, alongside 18 with the opportunity to more
easily control their budget and closely monitor their cart contents (KPMG International, 2012).
Thus, the aspect that seems to be, at least, one of the main drivers for online shoppers is the need
for convenient shopping activities. The study performed by Rohm & Swaminathan (2004), for
instance, confirms this result arguing that convenience, paired with variety seeking, are the main
underlying motivations in the online channel, while time savings and recreational shopping
orientations appear to be more significant drivers in the offline store format. Although
convenience remains important in both retail channels, one possible explanation the authors
found for the differentiating drivers between store formats may be that, while there is time saved
in shopping online, there is also a higher time gap between purchase and actual acquisition of the
goods purchased, due to the delivery waiting periods.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 9

 3. SOFTWARE AND HARDWARE REQUIREMENTS

3.1 Software Requirements

In order to develop this project, we need to install the following software/packages:

3.1.1 Anaconda Navigator

Anaconda Navigator is a free and open-source distribution of the Python and R programming
languages for data science and machine learning related applications.

It can be installed on Windows, Linux, and macOS. Anaconda is an open-source, cross-platform,

package management system. Anaconda comes with great tools like JupyterLab, Jupyter
Notebook, QtConsole, Spyder, Glueviz, Orange, Rstudio, Visual Studio Code. For this project,
we will be using Spyder.

In order to run, many scientific packages depend on specific versions of other packages. Data
scientists often use multiple versions of many packages and use multiple environments to
separate these different versions.

The command-line program anaconda is both a package manager and an environment manager.
This helps data scientists ensure that each version of each package has all the dependencies it
requires and works correctly.

Navigator is an easy, point-and-click way to work with packages and environments without
needing to type conda commands in a terminal window. You can use it to find the packages you
want, install them in an environment, run the packages, and update them – all inside Navigator.

The following applications are available by default in Navigator:

● JupyterLab

● Jupyter Notebook

● Spyder

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 10

 ● PyCharm

● VSCode

● Glueviz

● Orange 3 App

● RStudio

● Anaconda Prompt (Windows only)

● Anaconda PowerShell (Windows only)

How to install Anaconda Navigator

● Enter the url and download the latest version of anaconda navigator.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 11

 ● Click on Next to continue with the setup.

● Click on the I Agree option

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 12

 ● Select All Users option and click on Next.

● Click on Install button to start installing anaconda.

● Click on Finish after the installation.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 13

The simplest way to run code with navigator is with Spyder. From the Navigator Home tab, click
Spyder, and write and execute your code.

You can also use Jupyter Notebooks the same way. Jupyter Notebooks are an increasingly
popular system that combine your code, descriptive text, output, images, and interactive
interfaces into a single notebook file that is edited, viewed, and used in a web browser.

3.1.2 Jupyter Notebook

The Jupyter Notebook is an open-source web application that allows you to create and share
documents that contain live code, equations, visualizations and narrative text. Uses include: data
cleaning and transformation, numerical simulation, statistical modeling, data visualization,
machine learning, and much more.

Jupyter is a free, open-source, interactive web tool known as a computational notebook, which
researchers can use to combine software code, computational output, explanatory text and
multimedia resources in a single document. Computational notebooks have been around for
decades, but Jupyter in particular has exploded in popularity over the past couple of years. This
rapid uptake has been aided by an enthusiastic community of user–developers and a redesigned
architecture that allows the notebook to speak dozens of programming languages — a fact
reflected

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 14

in its name, which was inspired, according to co-founder Fernando Pérez, by the programming
languages Julia (Ju), Python (Py) and R.

3.1.3 Spyder

Spyder is a free and open-source scientific environment written in Python, for Python, and
designed by and for scientists, engineers and data analysts. It features a unique combination of
the advanced editing, analysis, debugging, and profiling functionality of a comprehensive
development tool with the data exploration, interactive execution, deep inspection, and beautiful
visualization capabilities of a scientific package.

Components

Editor

Work efficiently in a multilanguage editor with a function/class browser, code analysis tools,
automatic code completion, horizontal/vertical splitting and go-to-definition.

IPythonConsole

Harness the power of as many IPython consoles as you like in one GUI. Run code by line, cell or
file; or work interactively with debugging, plots and magic commands.

Variable Explorer

Interact with and modify variables on the fly: plot a histogram or time series, edit a date frame or
numpy array, sort a collection, dig into nested objects, and more.

Plots

Browse, zoom, copy and save the figures and images you create.

Debugger

Trace each step of your code’s execution interactively.

Help

Instantly view any object’s docs, and render your own.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 15

3.1.4 OpenCV (Open Source Computer Vision Library)

It is used for Image processing. OpenCV (Open Source Computer Vision Library) is a library of
programming functions mainly aimed at real-time computer vision. OpenCV-Python is a library
of Python bindings designed to solve computer vision problems. All the OpenCV array structures
are converted to and from NumPy arrays. This also makes it easier to integrate with other
libraries that use NumPy such as SciPy and Matplotlib.

Open Anaconda prompt and type the following command

● Type pip install OpenCV-python & click enter.

OpenCV (Open Source Computer Vision Library) is an open source computer vision and
machine learning software library. OpenCV was built to provide a common infrastructure for
computer vision applications and to accelerate the use of machine perception in the commercial
products. Being a BSD-licensed product, OpenCV makes it easy for businesses to utilize and
modify the code.

The library has more than 2500 optimized algorithms, which includes a comprehensive set of
both classic and state-of-the-art computer vision and machine learning algorithms. These
algorithms

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 16

can be used to detect and recognize faces, identify objects, classify human actions in videos,
track camera movements, track moving objects, extract 3D models of objects, produce 3D point
clouds from stereo cameras, stitch images together to produce a high resolution image of an
entire scene, find similar images from an image database, remove red eyes from images taken
using flash, follow eye movements, recognize scenery and establish markers to overlay it with
augmented reality, etc. OpenCV has more than 47 thousand users and an estimated number of
downloads exceeding 18 million. The library is used extensively in companies, research groups
and by governmental bodies.

Along with well-established companies like Google, Yahoo, Microsoft, Intel, IBM, Sony,
Honda, Toyota that employ the library, there are many startups such as Applied Minds,
VideoSurf, and Zeitera, that make extensive use of OpenCV. OpenCV’s deployed uses span the
range from stitching street view images together, detecting intrusions in surveillance video in
Israel, monitoring mine equipment in China, helping robots navigate and pick up objects at
Willow Garage, detection of swimming pool drowning accidents in Europe, running interactive art
in Spain and New York, checking runways for debris in Turkey, inspecting labels on products in
factories around the world on to rapid face detection in Japan.

It has C++, Python, Java and MATLAB interfaces and supports Windows, Linux, Android and
Mac OS. OpenCV leans mostly towards real-time vision applications and takes advantage of
MMX and SSE instructions when available. A full-featured CUDA and OpenCL interfaces are
being actively developed right now. There are over 500 algorithms and about 10 times as many
functions that compose or support those algorithms. OpenCV is written natively in C++ and has
a templated interface that works seamlessly with STL containers.

3.1.5 NumPy

It is an open-source numerical Python library. It contains a multidimensional array and matrix

data structures and can be used to perform mathematical operations. NumPy is a Python library
used for working with arrays. It also has functions for working in the domain of linear algebra,
fourier transform, and matrices. NumPy was created in 2005 by Travis Oliphant.

It is an open source project and you can use it freely. NumPy stands for Numerical Python.
NumPy offers comprehensive mathematical functions, random number generators, linear algebra

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 17

routines,

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 18

Fourier transforms, and more. NumPy supports a wide range of hardware and computing
platforms, and plays well with distributed, GPU, and sparse array libraries. The core of NumPy
is well-optimized C code.

NumPy is a Python library used for working with arrays. It also has functions for working in the
domain of linear algebra, fourier transform, and matrices. NumPy was created in 2005 by Travis
Oliphant. It is an open source project and you can use it freely.

Even for the delete operation, the NumPy array is faster. Because the NumPy array is densely
packed in memory due to its homogeneous type, it also frees the memory faster. So overall a task
executed in NumPy is around 5 to 100 times faster than the standard python list, which is a
significant leap in terms of speed.

Python is by far one of the easiest programming languages to use. NumPy is one such Python
library. NumPy is mainly used for data manipulation and processing in the form of arrays. It's
high speed coupled with easy to use functions make it a favorite among Data Science and
Machine Learning practitioners.

NumPy is a library for the Python programming language, adding support for large, multi-
dimensional arrays and matrices, along with a large collection of high-level mathematical
functions to operate on these arrays. Moreover, NumPy forms the foundation of the Machine
Learning stack.

Open Anaconda prompt and type the following command:

● Type “pip install numpy” and click enter.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 19

3.1.6 Keras

Keras is an open-source software library that provides a Python interface for artificial neural
networks. Keras acts as an interface for the TensorFlow library. Up until version 2.3 Keras
supported multiple backends, including TensorFlow, Microsoft Cognitive Toolkit, Theano, and
PlaidML.

Keras is a powerful and easy-to-use free open source Python library for developing and
evaluating deep learning models. It wraps the efficient numerical computation libraries Theano
and TensorFlow and allows you to define and train neural network models in just a few lines of
code.

Keras is a high-level neural networks API developed with a focus on enabling fast
experimentation. Being able to go from idea to result with the least possible delay is key to doing
good research. Keras has the following key features:

● Allows the same code to run on CPU or on GPU, seamlessly.

● User-friendly API which makes it easy to quickly prototype deep learning models.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 20

 ● Built-in support for convolutional networks (for computer vision), recurrent
networks (for sequence processing), and any combination of both.

● Supports arbitrary network architectures: multi-input or multi-output models,

layer sharing, model sharing, etc. This means that Keras is appropriate for
building essentially any deep learning model, from a memory network to a
neural Turing machine.

Keras contains numerous implementations of commonly used neural-network building blocks such
as layers, objectives, activation functions, optimizers, and a host of tools to make working with
image and text data easier to simplify the coding necessary for writing deep neural network code.
The code is hosted on GitHub, and community support forums include the GitHub issues page,
and a Slack channel.

In addition to standard neural networks, Keras has support for convolutional and recurrent neural
networks. It supports other common utility layers like dropout, batch normalization, and pooling.

Keras allows users to productize deep models on smartphones (iOS and Android), on the web, or
on the Java Virtual Machine. It also allows use of distributed training of deep-learning models on
clusters of Graphics processing units (GPU) and tensor processing units (TPU).

Keras is a minimalist Python library for deep learning that can run on top of Theano or
TensorFlow. It was developed to make implementing deep learning models as fast and easy as
possible for research and development.

Open Anaconda prompt and type the following command:

● Type “pip install keras==2.4.0” and click enter.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 21

3.1.7 TensorFlow

TensorFlow is an open-source library developed by Google primarily for deep learning

applications. It also supports traditional machine learning. TensorFlow was originally developed
for large numerical computations without keeping deep learning in mind.

How to download Tensorflow

1. Locate the path of Anaconda.

2. Set the working directory to Anaconda.

3. Create the yml file (For MacOS user, TensorFlow is installed here)

4. Edit the yml file.

5. Compile the yml file.

6. Activate Anaconda.

7. Install TensorFlow (Windows user only)

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 22

An open source framework for machine learning and other computations on decentralized data.
TensorFlow Probability is a library for probabilistic reasoning and statistical analysis. Tensor is a
library of deep learning models and datasets designed to make deep learning more accessible and
accelerate ML research.

TensorFlow is used to create large-scale neural networks with many layers. TensorFlow is
mainly used for deep learning or machine learning problems such as Classification, Perception,
Understanding, Discovering, Prediction and Creation.

Just like you might have done with Keras, it's time to build up your neural network, layer by
layer. If you haven't done so already, import tensorflow into your workspace under the
conventional alias tf. Then, you can initialize the Graph with the help of Graph(). You use this
function to define the computation.

TensorFlow makes it easy for beginners and experts to create machine learning models for
desktop, mobile, web, and cloud.

TensorFlow is a popular deep learning framework. You'll also preprocess your data: you'll learn
how to visualize your images as a matrix, reshape your data and rescale the images between 0
and 1 if required. With all of this done, you are ready to construct the deep neural network
model.

For researchers, Tensorflow is hard to learn and hard to use. Research is all about flexibility, and
lack of flexibility is baked into Tensorflow at a deep level. For machine learning practitioners
such as myself, Tensorflow is not a great choice either.

TensorFlow makes it easy for beginners and experts to create machine learning models. See the
sections below to get started. Guides explain the concepts and components of TensorFlow.

It's hard because it's very powerful and very complex. One thing I notice is a lot of ML tutorials
and resources more than any other field get outdated very quickly.

TensorFlow provides excellent functionalities and services when compared to other popular deep
learning frameworks. These high-level operations are essential for carrying out complex parallel
computations and for building advanced neural network models. TensorFlow is a low-level
library which provides more flexibility.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 23

Open Anaconda prompt and type the following command:

● Type “pip install tensorflow==2.3.0” and click enter.

3.1.8 IBM Cloud

What is the IBM Cloud platform?

The IBM Cloud platform combines platform as a service (PaaS) with infrastructure as a service
(IaaS) to provide an integrated experience. The platform scales and supports both small
development teams and organizations, and large enterprise businesses. Globally deployed across
data centers around the world, the solution you build on IBM Cloud spins up fast and performs
reliably in a tested and supported environment you can trust!

IBM Cloud provides solutions that enable higher levels of compliance, security, and
management, with proven architecture patterns and methods for rapid delivery for running
mission-critical workloads. Available in data centers worldwide, across 19 countries with multi
zone regions in North and South America, Europe, Asia, and Australia, you are enabled to
deploy locally with global scalability.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 24

IBM Cloud offers the most open and secure public cloud for business with a next-generation
hybrid cloud platform, advanced data and AI capabilities, and deep enterprise expertise across 20
industries. Solutions are available depending on your needs for working in the public cloud, on-
premises, or a combination:

● With a public cloud, the resources are made available to you over the public internet. It
is a multi-tenant environment, and resources like hardware and infrastructure are
managed by IBM.

● A hybrid cloud solution is a combination of public and private, giving you the flexibility
to move workloads between the two based on your business and technological needs.
IBM uses Red Hat OpenShift on IBM Cloud, the market-leading hybrid cloud container
platform for hybrid solutions that enables you to build once and deploy anywhere. With
IBM Cloud Satellite, you can create a hybrid environment that brings the scalability and
on-demand flexibility of public cloud services to the applications and data that runs in
your secure private cloud.

● Support for multi cloud and hybrid multi cloud solutions is also available, which makes
it easy for you to work with different vendors. IBM Cloud Paks are software products
for hybrid clouds that enable you to develop apps once and deploy them anywhere.

● Virtual Private Cloud (VPC) is available as a public cloud offering that lets you
establish your own private cloud-like computing environment on shared public cloud
infrastructure. With VPC, enterprises can define and control a virtual network that is
logically isolated from all other public cloud tenants, creating a private, secure place on
the public cloud.

With our open source technologies, such as Kubernetes, Red Hat OpenShift, and a full range of
compute options, including virtual machines, containers, bare metal, and serverless, you have the
control and flexibility that's required to support workloads in your hybrid environment. You can
deploy cloud-native apps while also ensuring workload portability.

Whether you need to migrate apps to the cloud, modernize your existing apps by using cloud
services, ensure data resiliency against regional failure, or use new paradigms and deployment

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 25

topologies to innovate and build your cloud-native apps, the platform's open architecture is built
to accommodate your use case.

What's built into the platform?

As the following diagram illustrates, the IBM Cloud platform is composed of multiple
components that work together to provide a consistent and dependable cloud experience.

● A robust console that serves as the front end for creating, viewing, managing your
cloud resources

● An identity and access management component that securely authenticates users for
both platform services and controls access to resources consistently across IBM
Cloud

● A catalog that consists of hundreds of supported products

● A search and tagging mechanism for filtering and identifying your resources

● An account and billing management system that provides exact usage for pricing
plans and secure credit card fraud protection.

Whether you have existing code that you want to modernize and bring to the cloud or you're
developing a brand new application, your developers can tap into the rapidly growing ecosystem
of available services and runtime frameworks in IBM Cloud.

Setting up your account

If you're a developer and you're just trying out IBM Cloud, you can go straight to the catalog and
browse the products that you'd like to explore and add to your Lite account. When you're ready
to get started with an environment and get apps running in production, consider setting up the
basics in your account:

● User access groups for organizing users and service IDs into one entity to make
assigning access a streamlined process.

● Resource groups for organizing your resources to make assigning access to a set of
resources quick and easy.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 26

 ● Access policies for your access groups or individual developers who need IAM
access policies or Cloud Foundry org and space roles.

As a financial officer for your company, you might be interested in simplifying how you manage
billing and usage across multiple teams and departments. With a Subscription account, you can
create an IBM Cloud enterprise, which offers centralized account management, consolidated
billing, and top-down usage reporting. An enterprise consists of an enterprise account, account
groups, and individual accounts.

● The enterprise account is the parent account to all other accounts in the enterprise.
Billing for the entire enterprise is managed at the enterprise account level.

● Account groups provide a way to organize related accounts. And, you get a unified
view of resource usage costs across all accounts that are included in an account
group.

● Similar to stand-alone accounts, accounts in an enterprise contain resources and

resource groups, Cloud Foundry orgs and spaces, and independent access
permissions.

IBM Cloud catalog

IBM Cloud provides a full-stack, public cloud platform with various products in the catalog,
including options for compute, storage, networking, end-to-end developer solutions for app
development, testing and deployment, security management services, traditional and open source
databases, and cloud-native services. You can find all of these services on the Services tab in the
catalog. The lifecycle and operations of these services are the responsibility of IBM.

The Software tab includes a growing catalog of software products, including Cloud Paks, starter
kits, Helm charts, Operators, and virtual server images. Even though you're responsible for the
lifecycle management, deployment, and configuration of these software products on your own
computer resources, you can take advantage of a simplified installation process to get up and
running quickly.

The catalog also supports command-line interfaces (CLIs) and a RESTful API for users to
retrieve information about existing products, and create, manage, and delete their resources.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 27

Searching the catalog

The following table lists the filter options that you can use when you search the catalog.

AI/Machine Learning: Products that enable systems to learn from data rather than through
explicit programming.

Analytics: Products that facilitate the analysis of data, typically large sets of business data, by the
use of mathematics, statistics, and other means.

Blockchain: Products that facilitate the process of recording transactions and tracking assets in a
business network.

Compute: Infrastructure resources that serve as the basis for building apps in the cloud.

Containers: A standard unit of software that packages up code and all its dependencies so the app
runs quickly and reliably from one computing environment to another.

Databases: Products that provide some form of access to a database without the need for setting
up physical hardware, installing software, or configuring for performance.

Developer Tools: Products that support developing, testing, and debugging software.

Integration: Products that facilitate the connection of data, apps, APIs, and devices across an
organization to be more efficient, productive, and agile.

Internet Of Things: Products that support receiving and transferring data over wireless networks
without human intervention.

Logging and Monitoring: Products that support storing, searching, analyzing, and monitoring log
data and events. And, products that support reviewing and managing the operational workflow
and processes being logged.

Mobile: Products with specific or special utility for users creatings things to be used on mobile
devices.

Networking: Products that support or augment the linking of computers so they can operate
interactively.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 28

Security: Products that provide the protection of stored data from theft, leakage, and deletion.

Storage: Products that support data to be created, read, updated, and deleted.

Pricing and billing

You can view the pricing details for each service when you're browsing the catalog. If you
choose a service plan with a paid plan, you can estimate your costs by using the cost estimator
tool.

IBM Cloud billing provides multiple services that ensure the IBM Cloud platform can securely
manage pricing, accounts, usage, and more.

Account management

Account management maintains the billing relationship with the customer. Each account is a
billing entity that represents a customer. This service controls account lifecycle, subscription,
user relationship, and organization.

Usage metering

With usage metering, service providers can submit metrics that are collected for resource
instances that are created by IBM Cloud users. Third-party service providers that deliver an
integrated billing service are required to submit usage for all active service instances every hour.

Usage reports

Usage reports return the summary for the account for the specified month. Account billing
managers are authorized to access the reports.

Managing security and compliance

The IBM Cloud Security and Compliance Center offers a single location where you can validate
that your resources are meeting continuous security and compliance.

You can create profiles and config rules to ensure that specific areas of your business adhere to
your defined requirements or industry regulations. From the Security and Compliance Center
dashboard, you can download detailed reports that you can use to provide evidence to

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 29

stakeholders

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 30

or external auditors. The Security and Compliance Center also offers security insights that you
can use to detect potential threats when observing your account activity.

Creating resources

The resource controller is the next-generation IBM Cloud platform provisioning layer that
manages the lifecycle of IBM Cloud resources in your account. Resources are created globally in
an account scope. The resource controller supports the creation of resources both synchronously
and asynchronously. Examples of resources include databases, accounts, processors, memory,
and storage limits.

In general, resources that are tracked by the provisioning layer are intended to associate usage
metrics and billing, but that isn’t always the case. In some cases, the resource might be
associated with the provisioning layer to ensure that its lifecycle can be managed along with the
account lifecycle. The resource controller uses IBM Cloud Identity and Access Management
(IAM) for authentication and authorization of actions that are taken against the provisioning
layer.

Resource lifecycle management

The resource controller provides common APIs to control the lifecycle of resources from
creating an instance to creating access credentials to removing access to deleting an instance.

Managing your resources

A collection of resources is managed by resource groups. A resource group is associated with

your account. All IBM Cloud resources must be assigned to a resource group. When you create
an account, a default resource group is created for you. All IBM Cloud IAM-enabled resources
0must be created within a resource group. If you have a Lite account, you can have only one
resource group. If you have a Pay-As-You-Go or Subscription account, you can create more than
one resource group. If an account is suspended, the corresponding resource group is suspended
as well, and all resources within the resource group are suspended.

Searching and tagging resources

The search service is a global and shared resource properties repository that is integrated within

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 31

the IBM Cloud platform. It is used for storing and searching a cloud resource's attributes, and it

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 32

categorizes and classifies resources. Resources are uniquely identified by a Cloud Resource
Name (CRN) identifier. The properties of a resource include tags and system properties. Both
properties are defined within an IBM Cloud billing account, and span across many regions.

This service also manages tags that are associated with a resource. You can create, delete, search,
attach, or detach tags with the Tagging API. Tags are uniquely identified by a CRN identifier.
Tags have a name, which must be unique within a billing account. You can create tags in
key:value pairs or label format.

Monitoring your resources

Observability offers a single location where you can monitor and observe your applications and
services in IBM Cloud.

With the IBM Log Analysis service, you can add log management capabilities to your IBM
Cloud architecture and you can manage system and application logs. It offers advanced features
to monitor and troubleshoot, define alerts, and design custom dashboards.

You can gain operational visibility into the performance and health of your applications,
services, and platforms with the IBM Cloud Monitoring service. It offers a full stack telemetry
with advanced features to monitor and troubleshoot, define alerts, and design custom dashboards.

Monitoring your account

Use the IBM Cloud Activity Tracker service to monitor the activity of your IBM Cloud account,
investigate abnormal activity and critical actions, and comply with regulatory audit requirements.
In addition, you can be alerted to actions as they happen. The events that are collected comply
with the Cloud Auditing Data Federation (CADF) standard.

Viewing status

The IBM Cloud Status page is the central place to find all unplanned incidents, planned
maintenance, announcements, and security bulletin notifications about key events that affect the
IBM Cloud platform. You can filter these categories by selecting specific locations, components,
types of ongoing events, or by using keyword searches.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 33

Notification preferences

Depending on your IBM Cloud account type, you can choose to receive email notifications about
IBM Cloud platform-related items and resource-related items from the Notification preferences
page. Platform-related items include announcements, billing and usage, and ordering. Resource-
related items include incidents, maintenance, security bulletins, and resource activity.

To complete this project you must have an IBM

account Step 1: Navigate to Cloud.Ibm.com

Step 2: Click on Create an account

● Fill the necessary information and Verify your email then click Create account.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 34

 ● Click on Login.

● Login using the credentials provided while registering and click on proceed.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 35

 ● Once you click on Proceed you will be redirected to the Dashboard. Hence with this step
you have successfully created an IBM account.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 36

Cloud Object Storage: IBM Cloud Object Storage makes it possible to store practically
limitless amounts of data, simply and cost effectively. It is commonly used for data archiving and
backup; for web and mobile applications; and as scalable, persistent storage for analytics.

Benefits

Reduce storage costs

Save on associated storage costs that include server, power, and data center space requirements.

Reduce downtime

Maintain more streamlined storage environments to help lower day-to-day touch points for IT
storage teams.

Strengthen developer team

Enable higher developer productivity from the increased agility in object-based storage
environments.

Create new business value

Increase scalability and performance with object-based storage environments to win more
business.

How is it used?

Backup and recovery

IBM Cloud Object Storage offers a scalable, secure destination to back up your critical data. It
reduces the cost of backups while still retaining immediate access.

Data archiving

Consolidate archive data and store it in IBM Cloud Object Storage, where it is cost-effective,
permanently available and protected.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 37

Cloud-native application

Build integrated apps using compute runtimes and microservices and IBM Cloud Object Storage
services for data storage. SDKs support API functions.

AI and big data analytics

Create a data lake in IBM Cloud Object Storage and extract actionable insights, using query-in-
place, analytics and machine-learning tools.

Features

Smart Tier

Smart Tier automates tier classification and cost optimization based on data activity. It’s ideal for
unknown or changing data usage patterns.

High durability and resilience

Built for 99.99999999% data durability. Select the resiliency option for the location, availability
and performance you need. Individual results vary.

Data protection and security

Manage encryption keys with IBM Key Protect. Set role-based policies and access permission
with IBM Cloud Identity and Access Management Services.

Fast data transfer

Securely move data to IBM Cloud Object Storage with the natively integrated Aspera high-speed
data transfer option. Upload data at no cost.

Query data in place

Analyze data fast in IBM Cloud Object Storage with IBM SQL Query. Tap into your data simply
to extract, transform and load (ETL) data.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 38

Immutable object storage

Preserve electronic records and protect data against deletion or modification.

Quota management

Manage storage cost by limiting bucket usage. Once the limit is reached, you can’t add data
unless you extend the limit. Set a threshold-based alert.

Keep your own key

Integration with IBM Cloud Hyper Protect Crypto Services offers enhanced data security,
encryption options and more granular control and authority

Accelerated archive

With a faster restore option of up to two hours, you can access your dormant data faster, while
saving on storage costs for your long-term data.

CloudantDB

A fully managed, distributed database optimized for heavy workloads and fast-growing web and
mobile apps, IBM Cloudant is available as an IBM Cloud service with a 99.99% SLA. Cloudant
elastically scales throughput and storage, and its API and replication protocols are compatible
with Apache CouchDB for hybrid or multi cloud architectures.

How is it used?

Serverless web app and API

Here, we will create a serverless web application by hosting static website content on github
pages and implementing the application back end, using IBM Cloud Functions.

Mobile app with serverless

In this tutorial, you'll learn how to use IBM Cloud Functions along with cognitive and data
services to build a serverless back end for a mobile app.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 39

Find anomalies in IoT data

This tutorial shows how to set up an IoT device and gather data in the IBM Watson IoT
Platform. Create visualizations and use advanced ML services to analyze historical data and
detect anomalies.

Open hybrid multi cloud

It shows how to pair the API and powerful replication protocol of cloudant with Apache
CouchDB in a hybrid cloud environment.

Features

Serverless

Instantly deploy an instance, create databases and independently scale throughput capacity and
data storage to meet your application requirements.

Secure

Encrypt all data, with optional user-defined encryption key management through IBM Key
Protect, and integrate with IBM Identity and Access Management.

Global availability

Get continuous availability as cloudant distributes data across availability zones and 6 regions
for app performance and disaster recovery requirements.

3.1.9 Node-RED

Node-RED is a programming tool for wiring together hardware devices, APIs and online
services in new and interesting ways.

It provides a browser-based editor that makes it easy to wire together flows using the wide range
of nodes in the palette that can be deployed to its runtime in a single click.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 40

Features

Browser-based flow editing

Node-RED provides a browser-based flow editor that makes it easy to wire together flows using
the wide range of nodes in the palette. Flows can be then deployed to the runtime in a single-
click.

JavaScript functions can be created within the editor using a rich text editor.

A built-in library allows you to save useful functions, templates or flows for re-use.

Built on Node.js

The light-weight runtime is built on Node.js, taking full advantage of its event-driven, non-
blocking model. This makes it ideal to run at the edge of the network on low-cost hardware such
as the Raspberry Pi as well as in the cloud.

With over 225,000 modules in Node's package repository, it is easy to extend the range of palette
nodes to add new capabilities.

Social Development

The flows created in Node-RED are stored using JSON which can be easily imported and
exported for sharing with others.

An online flow library allows you to share your best flows with the world.

3.2 Hardware Requirements

Processor : Two Quad Core

RAM : 8GB

Hard Drive : 5 x146GB RAID

NIC : 4 x10/100mb

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 41

 4. CODING AND IMPLEMENTATION

Module: tensorflow.keras.preprocessing.image

Set of tools for real-time data augmentation on image data.

Classes

class DirectoryIterator: Iterator capable of reading images from a directory on disk.

Class ImageDataGenerator: Generate batches of tensor image data with real-time data
augmentation.

class Iterator: Base class for image data iterators.

class NumpyArrayIterator: Iterator yielding data from a Numpy array.

Functions

apply_affine_transform(...): Applies an affine transformation specified by the parameters given.

apply_brightness_shift(...): Performs a brightness shift.

apply_channel_shift(...): Performs a channel shift.

array_to_img(...): Converts a 3D Numpy array to a PIL Image instance.

img_to_array(...): Converts a PIL Image instance to a Numpy array.

load_img(...): Loads an image into PIL format.

random_brightness(...): Performs a random brightness shift.

random_channel_shift(...): Performs a random channel shift.

random_rotation(...): Performs a random rotation of a Numpy image tensor.

random_shear(...): Performs a random spatial shear of a Numpy image tensor.

random_shift(...): Performs a random spatial shift of a Numpy image tensor.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 42

random_zoom(...): Performs a random spatial zoom of a Numpy image tensor.

save_img(...): Saves an image stored as a Numpy array to a path or file object.

smart_resize(...): Resize images to a target size without aspect ratio distortion.

4.1 Sample Code

From tensorflow.keras.preprocessing.image we are importing the ImageDataGenerator class.

We are applying the function ImageDataGenerator to the test and training datasets.

Rescale: rescaling factor. Defaults to None. If None or 0, no rescaling is applied, otherwise we

multiply the data by the value provided (after applying all other transformations).

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 43

Zoom_range: Float or [lower, upper]. Range for random zoom. If a float, [lower, upper] = [1-
zoom_range, 1+zoom_range].

horizontal_flip: Boolean. Randomly flip inputs horizontally.

vertical_flip: Boolean. Randomly flip inputs vertically.

Width_shift_range: Float, 1-D array-like or int

● float: fraction of total width, if < 1, or pixels if >= 1.

● 1-D array-like: random elements from the array.

● int: integer number of pixels from interval (-width_shift_range, +width_shift_range)

● With width_shift_range=2 possible values are integers [-1, 0, +1], same as with
width_shift_range=[-1, 0, +1], while with width_shift_range=1.0 possible values are
floats in the interval [-1.0, +1.0).

We are giving the path of both test and train datasets. After this, it will display the number of
images and classes.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 44

Module:tensorflow.keras.models

Model groups layers into an object with training and inference features.

Sequential Model

A Sequential model is appropriate for a plain stack of layers where each layer has exactly one
input tensor and one output tensor.

A Sequential model is not appropriate when:

● Your model has multiple inputs or multiple outputs

● Any of your layers has multiple inputs or multiple outputs

● You need to do layer sharing

● You want non-linear topology (e.g. a residual connection, a multi-branch model)

Importing Sequential model from tensorflow.keras.models module.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 45

Module tensorflow.keras.layers

Class Convolution2D

It is a 2D convolution layer (e.g. spatial convolution over images).

Inherits From: Layer, Module

This layer creates a convolution kernel that is convolved with the layer input to produce a tensor
of outputs. If use_bias is True, a bias vector is created and added to the outputs. Finally, if
activation is not None, it is applied to the outputs as well.

When using this layer as the first layer in a model, provide the keyword argument input_shape
(tuple of integers or None, does not include the sample axis), e.g. input_shape=(128, 128, 3) for
128x128 RGB pictures in data_format="channels_last". You can use None when a dimension has
variable size.

Class MaxPooling2D

Max pooling operation for 3D data (spatial or spatio-temporal).

Inherits From: Layer, Module

It Downsamples the input along its spatial dimensions (depth, height, and width) by taking the
maximum value over an input window (of size defined by pool_size) for each channel of the
input. The window is shifted by strides along each dimension.

Input shape:

● If data_format='channels_last': 5D tensor with shape: (batch_size, spatial_dim1,

spatial_dim2, spatial_dim3, channels)

● If data_format='channels_first': 5D tensor with shape: (batch_size, channels,

spatial_dim1, spatial_dim2, spatial_dim3)

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 46

Output shape:

● If data_format='channels_last': 5D tensor with shape: (batch_size, pooled_dim1,

pooled_dim2, pooled_dim3, channels)

● If data_format='channels_first': 5D tensor with shape: (batch_size, channels,

pooled_dim1, pooled_dim2, pooled_dim3)

Class Flatten

Flattens the input. Does not affect the batch size.

Inherits From: Layer, Module

If inputs are shaped (batch,) without a feature axis, then flattening adds an extra channel
dimension and output shape is (batch, 1).

Class Dense

Just your regular densely-connected NN layer.

Inherits From: Layer, Module

Dense implements the operation: output = activation(dot(input, kernel) + bias) where activation is the
element-wise activation function passed as the activation argument, kernel is a weights matrix
created by the layer, and bias is a bias vector created by the layer (only applicable if use_bias is
True). These are all attributes of Dense.

If the input to the layer has a rank greater than 2, then Dense computes the dot product between
the inputs and the kernel along the last axis of the inputs and axis 1 of the kernel (using
tf.tensordot). For example, if input has dimensions (batch_size, d0, d1), then we create a kernel
with shape (d1, units), and the kernel operates along axis 2 of the input, on every sub-tensor of
shape (1, 1, d1) (there are batch_size * d0 such sub-tensors). The output in this case will have
shape (batch_size, d0, units).

Besides, layer attributes cannot be modified after the layer has been called once (except the
trainable attribute). When a popular kwarg input_shape is passed, then keras will create an input

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 47

layer to insert before the current layer. This can be treated equivalent to explicitly defining an
InputLayer.

Importing Convolution2D, MaxPooling2D, Flatten, Dense classes from tensorflow.keras.layers

module.

● Classifier is a model name, it can be anything and we are initialising it.

● We are adding the convolution layer and giving the dimensions for the input image.
Activation ‘relu’ is used for regression .

● We are adding the MaxPooling layer, and giving the pool size i.e., size of the maximum
extracted features.

● We are doing the above two steps for two hidden layers.

● Then using the Flatten class we are converting the input into one dimensional.

● Next we are adding the dense layer. Dense is used to add the input layer and for
initialising the kernel. We are using two hidden layers and one output layer.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 48

 ● The output generated by the dense layer is an 'm' dimensional vector. Thus, a dense layer
is basically used for changing the dimensions of the vector. Dense layers also apply
operations like rotation, scaling, translation on the vector. Then we are getting the output
shape.

● Units: Number of inputs. Initializers define the way to set the initial random weights of
Keras layers.

● In the output layer, we are giving ‘softmax’ as input to the activation.

● The softmax function is used as the activation function in the output layer of neural
network models that predict a multinomial probability distribution.

● Softmax is used as the activation function for multi-class classification problems where
class membership is required on more than two class labels.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 49

 ● Using summary class, we summarize everything we built till now.

● The summary is textual and includes information about:

● The layers and their order in the model.

● The output shape of each layer.

● The number of parameters (weights) in each layer.

● The total number of parameters (weights) in the model.

Here, we are compiling the model using the compile function. The compilation is the final step in
creating a model.

Loss function is used to find error or deviation in the learning process. Keras requires a loss
function during the model compilation process.

Keras provides quite a few loss function in the losses module and they are as follows −

● mean_squared_error

● mean_absolute_error

● mean_absolute_percentage_error

● mean_squared_logarithmic_error

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 50

 ● squared_hinge

● hinge

● categorical_hinge

● log cosh

● huber_loss

● categorical_crossentropy

● sparse_categorical_crossentropy

● binary_crossentropy

● kullback_leibler_divergence

● poisson

● cosine_proximity

● is_categorical_crossentropy

All above loss function accepts two arguments −

● y_true − true labels as tensors

● y_pred − prediction with same shape as y_true

Optimization is an important process which optimizes the input weights by comparing the
prediction and the loss function.

Metrics is used to evaluate the performance of your model. It is similar to loss function, but not
used in the training process. Keras provides quite a few metrics as a module, metrics and they
are as follows

● accuracy

● binary_accuracy

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 51

 ● categorical_accuracy

● sparse_categorical_accuracy

● top_k_categorical_accuracy

● sparse_top_k_categorical_accuracy

● cosine_proximity

● clone_metric

Similar to loss function, metrics also accepts below two arguments −

● y_true − true labels as tensors

● y_pred − prediction with same shape as y_true

fit_generator is used when either we have a huge dataset to fit into our memory or when data
augmentation needs to be applied. Here we start by first initializing the number of epochs we are
going to train our network for along with the batch size.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 52

We then initialize aug, a Keras ImageDataGenerator object that is used to apply data
augmentation, randomly translating, rotating, resizing, etc. images on the fly.

Performing data augmentation is a form of regularization, enabling our model to generalize better.

However, applying data augmentation implies that our training data is no longer “static” — the
data is constantly changing.

Each new batch of data is randomly adjusted according to the parameters supplied to
ImageDataGenerator .

Thus, we now need to utilize Keras’ fit_generator function to train our model.

As the name suggests, the fit_generator function assumes there is an underlying function that is
generating the data for it. The function itself is a Python generator.

Internally, Keras is using the following process when training a model with fit_generator :

1. Keras calls the generator function supplied to fit_generator (in this case, aug.flow ).

2. The generator function yields a batch of size BS to the fit_generator function.

3. The fit_generator function accepts the batch of data, performs backpropagation, and
updates the weights in our model.

4. This process is repeated until we have reached the desired number of epochs.

5. You’ll notice we now need to supply a steps_per_epoch parameter when calling

fit_generator (the .fit method had no such parameter).

We compute the steps_per_epoch value as the total number of training data points divided by the
batch size. Once Keras hits this step count it knows that it’s a new epoch.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 53

We are saving our model by naming it as ‘gesture’.

Convert the object to a JSON string.

NaN’s and None will be converted to null and datetime objects will be converted to UNIX
timestamps.

Here, we are loading our gesture model and testing our model by giving the image in our test set.

And we are predicting the class.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 54

Creating Customer Database in cloudant

In this, we will be creating a cloud object storage(COS ) bucket and Customer Cloudant database

● COS Bucket - To store the detected faces

● Cloudant DB - To store URL of Detected images as well as store customer details

Create Cloud object Storage Bucket

Let's create a Cloud object Storage Bucket .

Upon login to IBM Cloud, you will be redirected to IBM Dashboard.

From the dashboard Click on Storage.

Once you login to IBM click on Storage, open Cloud object storage service.

Click on the service available .

You get navigated to the Object storage dashboard.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 55

Select Buckets and click on Create Bucket. A pop up display selects Custom Bucket.

Configure the bucket details as shown in the below figure

Bucket name: usersimages

Resiliency: Regional

Location: jp-tok

Storage Class: standard

Scroll down and click on create.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 56

Your Bucket is created.

Change Policy of Bucket

Now that we have created the bucket, let's give the public access to the bucket.

Click on Access Policies.

Select Public Access.

Click on Create Policy.

A pop-up displays click on Enable.

Close the popup window.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 57

Now Click on Buckets you will see the access of Bucket is Public.

Grab Cloud Object Storage End point

Once the bucket is created we can upload any documents or images or video files to the bucket
either through the console or through programming. In order to Upload documents through
Programing you should need credentials as follows:

● API Key

● Authentication endpoint

● COS CRN

Authentication Endpoint :

If you have remembered we have created our Bucket in jp-tok Region, each bucket region will
have a different endpoint. you get this endpoint from the endpoints option.

Click on Endpoint select the region and Resiliency.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 58

Copy the End Point - we need this in a python script that we going to make in the next activities

S3.jp-tok.cloud-object-storage.appdomain.cloud

Grab Cloud Object Storage Service Credentials

Now let's create service credentials to save the other two necessary creds.

Select Service Credentials and click on New Credential.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 59

Give name as Grocery and click on ADD.

Click on grocery and save the marked cred for future purposes.

"apikey": "1-xLp8tyZtKw9owdRdlGBHorT9n98AmVr8SiN2-SJMsL",

"endpoints": "https://control.cloud-object-storage.cloud.ibm.com/v2/endpoints",

"iam_apikey_description": "Auto-generated for key d962f883-f14f-4c11-86b9-a3e8e242627b",

"iam_apikey_name": "Grocery",

"iam_role_crn": "crn:v1:bluemix:public:iam::::serviceRole:Writer",

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 60

"iam_serviceid_crn": "crn:v1:bluemix:public:iam-
identity::a/8956b3127dfe48a5aa1b2fab1e8bf1bf::serviceid:ServiceId-4b8648bb-26c2-4e06-
8719-8b8753a016f6",

"resource_instance_id": "crn:v1:bluemix:public:cloud-object-
storage:global:a/8956b3127dfe48a5aa1b2fab1e8bf1bf:2d1c5aa1-5459-4ab9-9076-
f609ef602407::"

Create Customer database

In order to store the image URL and details of customers, you have to create Databases.

We will be creating a database to store image URLs using programming, but let us create a
customer database to store details manually.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 61

The below image shows the format of each customer. It contains a unique id, key, items
purchased and other details.

Create Product Offers Database

In order to store the product’s offers, you need to create a product offers database.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 62

The format of the product document is shown below. Here we have specified the discount for
each product.

Save Cloudant DB Credentials

As we are going to save the Image URL through programming we need some credentials which
are to be configured in the python script. Lets save the credentials from IBM Dashboard click on
services.

Click on Cloudant Database Service.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 63

Click on service credentials and open the available service credentials.

Copy and save username password and url.

"username": "apikey-v2-23wv2y7u5lm5jsd32ecfdg8o4ij0pfw4g02q0wxre86i"

"password": "7d45bd6c69660ea483a269aafc26084a"

"url": "https://apikey-v2-
23wv2y7u5lm5jsd32ecfdg8o4ij0pfw4g02q0wxre86i:7d45bd6c69660ea483a269aafc26084a@b6
e4b265-79c5-47ca-ae29-b0013433eda4-bluemix.cloudantnosqldb.appdomain.cloud"

OpenCV Analysis

Let's build a python script that captures the video frames, detects the faces in video frames and
stores the image in Cloud object storage. The image url is also stored in Cloudant DB.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 64

Import necessary libraries

Import the Following Libraries in your script.

Initialize haarcascade file paths - refer to the downloaded project file.

Start the video capturing using the below Code.

0 - for internal Camera

1 for external camera

Initialise Cloud Object Storage and Cloudant Credentials

initialize the creds like API Key Authentication endpoint Cos CRN etc as shown in the below
image .

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 65

navigate to cloud object storage service to get COS_API_KEY_ID and
COS_RESOURCE_CRN(resource instance_ID)

We have saved these credentials in the previous milestone.

Copy the API key and paste it into the code at COS_API_KEY_ID

Copy resource instance id and paste near COS_RESOURCE_CRN

Copy end Point and paste near COS_AUTH ENDPOINT

Initialize Cloudant credentials Copy user name password and URL and paste it in the below code

Initialize Necessary variables

● database_name - Database name which stores the images URL of detected Face

● picname : name of the image that gets stored in COS

● pic: name of the image that gets appended to image the URL

Cos - Establishing the connection to cloud object storage service using given credentials.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 66

Define a Function to Store image in Cloud object storage

let us create a function that stores image in the cloud object storage bucket

We are Configuring the image properties of images that will be stored in COS.

Save the Detected Face

Once all the necessary initializations are done, let's write the script for storing the image in Cos,
and parlelley lets create a URL for the detected image and store it in cloud DB.

Loop over the block to first read the frames and apply face detection algorithm as shown in the
below image

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 67

Save each face image detected in the frame in the local directory with the image name we have
initialized before.

Showcase the detected image on the open cv window with a rectangle drawn around the face and
call the defined function to store the image in the bucket created. In our case it is userimages.

In the below code, we are checking whether we have the database or if not it will create the
database with the name given (customer images).

Cloudant BD accepts JSON documents so we are creating a JSON document which has the id
and URL of the image.

Once the url is created we are storing the JSON document in the created database.

You can terminate the video capturing by pressing q on the window while the program is

running. The video capturing stops, the camera gets released and the open cv window gets

closed.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 68

The client gets disconnected from the cloud.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 69

Test the Output

Now that you have finished writing the code, let’s run the code and see whether the

● detected face images are getting saved in the bucket

● detected image URL is getting saved in the CloudantDB.

Run the code

Another window opens and you can see the live stream of the video and the detected face.

Navigate to the bucket created in Cos and see the saved images.

Navigate to cloudant DB, open the customer images database and see the url is stored.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 70

Build Node Red Application

Now let us create a web application where

● New users can get registered

● The recommendations for the recognized person is Displayed

Create a Node-red flow which has the following nodes

● Drag an HTTP request node on to the flow which fetches the mage URL from Cloudant
DB

● Parse the Image from image URL using JSON node

● Slow case the image on template node

● Give the Image to the visual recognition node to recognize the customer's name

● Grab the previous purchase history of the customer based on recognised name

● Showcase the Offers on UI

Fetch the image from Cloudant DB

Now let's create a flow that fetches Image from Cloudant DB:

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 71

Search for inject node from left palate, drag inject node and on to the flow and configure it as
shown below.

For every five seconds the inject node is triggered.

Drag HTTP request node and configure it with Cloudant DB credentials.

Select type as basic authentication.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 72

Give cloudant db username and password.

Get the URL from the customerimages database.

Navigate to the customer image database.

Click on JSON option as shown in the below image

It will open in a new browser, copy the browser link.

Add the following text to the copied LINK

?include_docs=true&descending=true&limit=1

Which looks similar to below link

https://83d0fbd2-6007-46f4-bf6e-e34b4edbb4bb-
bluemix.cloudant.com/customerimages/_all_docs?include_docs=true&descending=true&limit=1

Display the Fetched image on UI

Now let's drag the template node Configure the Node.

Assign Group and Tab.

Assign image URL as shown below

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 73

Code in template node.

<img src= {{msg.payload.rows[0].doc.link}}

alt="entered person" width="500" height="600">

Deploy the app and see the dashboard.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 74

Recommend the products

Now give the predicted name to the databases created.

Fetch the previous history products.

Fetch the offers in the products.

Display the offers on the dashboard.

Open the dashboard and output looks like this.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 75

4.2 Data Dictionary

The dataset is divided into train set and test set. Train set contains four classes and each class
contains 160 images. Test set contains four classes and each class contains 40 images.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 76

 5. OUTPUT SCREENS

This is how the output looks. In the dashboard, the image of the customer along with the
customer details and the product offers are displayed.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 77

 6. CONCLUSION

The “Grocery Offer Suggestion System for Retail Markets” project has been successfully
completed. The goal of the system is achieved, and the problem is solved.

Grocery offer system will provide appropriate product offer suggestions to the customers. By
using the previous purchased history, we provided the offers to the customers in a physical
grocery store. This will induce the customers to purchase products in the retail market, which
will also boost the business of the grocery store.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 78

 7. FUTURE ENHANCEMENTS

As seen earlier, recommendations are suggestions or lists of items that users might like and these
recommendations are independent of the consumer. The field is continually evolving and
changing.

To a business, optimizing their recommender is directly proportional to maximizing their

revenue. Therefore, we are trying to implement the proposed approach in the recommender
systems to give personalized recommendations.

This helps in creation of custom alternatives that meet the individual customer’s preferences
based on the purchase history and the domain knowledge.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 79

 8. BIBLIOGRAPHY AND REFERENCES

Pan, S.; Giannikas, V.; Han, Y.; Grover-Silva, E.; Qiao, B. Using customer-related data to
enhance e-grocery home delivery. Ind. Manag. Data Syst. 2017,117, 1917–1933.

Darley, W.K.; Blankson, C.; Luethge, D.J. Toward an integrated framework for online consumer
behavior and decision making process: A review. Psychol. Mark. 2010,27, 94–116.

Engel, J.F.; Blackwell, R.D.; Miniard, P.W. Consumer Behavior, 5th ed.; Dryden: Hinsdale, IL,
USA, 1986.

Nguyen, D.H.; de Leeuw, S.; Dullaert, W.E. Consumer behaviour and order fulfilment in online
retailing: A systematic review. Int. J. Manag. Rev. 2018,20, 255–276.

Rotem-Mindali, O.C.; Salomon, I. Modeling consumers’ purchase and delivery choices in the
face of the information age. Environ. Plan. B Plan. Des.

GROCERY OFFER SUGGESTION SYSTEM FOR RETAIL MARKETS 80