You are on page 1of 45

E-Zone Cart and

Recommendation System
Name: Deepak. S(MSC SS)
Register Number:1631010
Project Guide: Mrs.K.Rajarajeshwari
Company Name :DomainHostly
PROJECT – I

E-ZONE CART
PROJECT OBJECTIVE
• The primary goal of this project is to develop an e-commerce store where products can be
bought from the comfort of home through the Internet. Payments can be done online and the
products will be delivered to the customer’s home.

• This project deals with developing an e-commerce website for online product sale. It provides
the user with a catalogue of different products available to purchase in the store. In order to
facilitate online purchase a shopping cart is provided to the user. The system is implemented
using a three-tier approach with a backend database, a middle tier of PHP Laravel framework
and web browser as the front end client.
USE CASE DIAGRAM
ARCHITECTURAL DESIGN
MODULE DIAGRAM
ACTIVITY DIAGRAM
PRODUCT PAGE
• Register user can view the product’s full screen images.

• User can see the product’s specification, review, comments

• User can select the quantity of the product.

• User can add the product to the cart.


ORDER VIEW PAGE
• It displays all the products user wants to buy.

• User can have a grant total of the all items.

• After confirmation user can place the order from here.


PAYMENT GATEWAY
• After ordering the items the user is redirected to payment page.

• User can make the payment via credit/debit cards, Netbanking, UPI.

• The payment gateway used is Razorpay.


RAZORPAY

• Razorpay uses tokenisation to prevent data exposure

• Razorpay replaces your 16-digit card number with a single token, which is a unique set of

characters that replace your original card number. 

• Tokens are assigned randomly, which makes it impossible to reverse-the actual card number.

• Payments are processed in ease without the risk of data theft. 


TRACK ORDER
• After successful order placement an order ID is generated.

• The order id is sent to the user via email.

• The user can track the order using the order ID.
UI DESIGN
UI DESIGN
UI DESIGN
UI DESIGN
DATA MANAGEMENT
• This database consists of

• Users: User and Admin information is added to database with Unique ID based on their roles.

• Shopping: Complete products information is stored in this table.

• Orders: Customer ordered products, status and delivery information is stored in this table.
DATABASE TABLE DIAGRAM
ADMIN PAGE
ORDER PAGE
PAYMENT INFORMATION
CONCLUSION
• This application is successfully devised and implemented using PHP Laravel framework. This website help
users to view all the product information and to purchase the products It also gives a platform to post blogs
periodically.

• The users can add the product to the cart and can make the payment in future. The payment details along with
product purchase confirmation will be triggered as a mail to the prospect. The website is completely managed
with high security.

FUTURE ENHANCEMENTS

In future, the application will be enhanced by implementing with multiple payment gateways for user
convenience, auto refund feature for the failed transactions, OAuth2.0 for easy authentication, CDN servers can
be implemented to minimize delays in loading web pages content.
PROJECT – II

RECOMMENDATION SYSTEM
RECOMMENDATION SYSTEM
• Recommendation systems are machine learning systems that help users discover new product and
services.

• The objective of the project is to build a recommendation system  that help users discover new
product and services. Recommendation systems are designed to predict users interests and
recommend items a user might be interested in.

• The data can be obtained implicitly or explicitly. Acquiring the data implicitly by keeping track of
users’ behavior, such as tracking web sites visited or products purchased, while explicit data is the
result of collecting user ratings
USE CASE DIAGRAM
ACTIVITY DIAGRAM
COLLABORATIVE FILTERING ALGORITHM
• Collaborative filtering is a method that filter items based on recommendations of others.

• Collaborative filtering bases its recommendations and predictions either on ratings or behaviour of
other users by gathering data from other users’ opinions.

• Based on the data it gathers, it can make a reasonable prediction of one’s preference.
TYPES

1. Memory Based Collaborative Filtering

2. Model Based Collaborative Filtering


MEMORY BASED COLLABORATIVE FILTERING

• Memory-based collaborative filtering utilizes the entire user-item data to generate predictions. The
system uses statistical methods to search for a set of users who have similar transactions history to
the active user

• In the memory-based approach, all ratings are stored as-is into memory , and a recommendation
for the test user can be generated by sorting the memorized ratings of similar users or items
MEMORY BASED COLLABORATIVE FILTERING

USER-ITEM INTERACTION MATRIX

• An user-item interaction matrix is formed for all users with the items they rated. To find a users
rating for a particular product that is not been rated the similar users are found by their ratings and
the average of their rating is considered to be the rating of the non-rated user.

• To recommend products the distance between the n similar users are calculated by Euclidean
distance and the score for the product is predicted and it is recommended to the particular user.

• This makes it a large process as if there are many users and items they rated these matrix will be
huge and the prediction will be not accurate.
MODEL BASED COLLABORATIVE FILTERING

• In the model-based approach, training ratings are used to generate a model that is able to predict
the ratings for items that a test user has not rated before.

• The model-based methods have the resulting ‘compact’ models to solve the data sparsity problem
to a certain extent. It separate the data into training set and test set.

• To develop a model, there are two approaches that can be used, which are probability approach or
rating prediction. The modeling process is conducted by machine learning techniques such as
classification, clustering approach
WHY MODEL BASED IS BETTER THAN MEMORY BASED

• In term of accuracy of recommendation the model-based accuracy is better than memory-based


accuracy.

• Based on the computation time, model-based has an average computation time 10 times faster than
memory-based. This makes model-based better than memory based in terms of computational
speed in recommending products

• Based on the average number of relevant products perceived by user, model-based is better than
memory-based in generating relevant recommendation.
PRODUCT POPULARITY
• This system is mainly targeted for new customers. When a new customer without any previous
purchase history visits the e-commerce website for the first time, he/she is recommended the
most popular products sold on the company's website.

• The test data is taken from Kaggle(Amazon products ratings) and the most popular products
are filtered using the ratings given to the product. A graph is also plotted according to the data.
OUTPUT

Dataset that has been imported from Kaggle to


build the recommendation system
Popular products in the dataset. They are filtered
as popular products based on the ratings given to
them.
The graph gives us the most popular products (arranged in descending order)
sold by the business.
MODEL BASED COLLABORATIVE FILTERING
FOR E-COMMERCE BUSINESS
A model based collaborative filtering technique is chosen here as it helps in
making prediction of products for a particular user by identifying patterns based
on preferences from multiple user data.

UTILITY MATRIX

An utility matrix is consists of all possible user-item preferences (ratings)


details represented as a matrix. The utility matrix is sparce as none of the users
would buy all the items in the list, hence, most of the values are unknown.
OUTPUT

As expected, the utility matrix obtained above is sparce, the unknown


values are filled up with 0.
MODEL BASED COLLABORATIVE FILTERING FOR E-COMMERCE
BUSINESS

MATRIX DECOMPOSITION:

• The decomposition of matrix is done by using a technique called singular value decomposition.

• Singular Value Decomposition (SVD) is based on the matrix factorization (MF) technique that
utilizes a Latent factor model

• For example, if x number of users like a product Y and if they also like a product Z, then the
Latent factor model will group them together in the same Latent factor space Therefore, making it
possible to compare two users by evaluating their ratings for different features instead of
individual products.
MODEL BASED COLLABORATIVE FILTERING FOR E-COMMERCE
BUSINESS
MATRIX CORRELATION

• The decomposed matrix is correlated and Correlation for all items are considered and based on
the item purchased by this customer , based on items rated by other customers who bought the
same product.

• Then the top 10 highly correlated products are isolated in sequence and they are recommended to
the users.
OUTPUT

These top 10 products are displayed by the recommendation system to the


customer based on the purchase history of other customers in the website .
ITEM BASED FILTERING
•When a new product is fetched into a website it does not have any ratings
or review. Hence based on the description given to the products this
recommendation is done. 

•Converting the text in product description into numerical data for analysis.
product clusters for subset of data are created and visualized. Top words in each
cluster are taken based on the product description. Based on the users key
search words the cluster is predicted.

•Once a cluster is identified based on the user's search words, the


recommendation system can display items from the corresponding product
clusters based on the product description.
In case a word appears in multiple clusters, the algorithm chooses
the cluster with the highest frequency of occurrence of the word.
CONCLUSION
• The system is successfully implemented for recommending products to the user based on their interest. This works best
if a business is setting up its e-commerce website for the first time and does not have user-item purchase/rating history to
start with initially.

• This recommendation system will help the users get a good recommendation to start with and once the buyers have a
purchased history, the recommendation engine can use the model based collaborative filtering technique.

FUTURE ENHANCEMENTS

• The system can be improvised by using cosine similarity. Cosine Similarity is a metric used to measure how similar the
two items or documents are irrespective of their size.

• It measures the cosine of an angle between two vectors projected in multi-dimensional space. Thus the scalability and
the efficiency of the system can be improved.
REFERENCES

• P. H. Aditya, I. Budi, Q. Munajat, A Comparative Analysis of Memory-based and Model-based


Collaborative Filtering on the Implementation of Recommender System for Ecommerce in Indonesia,
2016.

• F. McCarey, M. Cinneide, & N. Kushmerick, A Recommender Agent for Software Libraries: An Evaluation
of Memory-Based and ModelBased Collaborative Filtering Proceedings of the IEEE/WIC/ACM
international conference on Intelligent Agent Technology, 2006.

• J. Schafer, J. Konstan, & J. Riedl, Recommender Systems in ECommerce Proceedings of the 1st ACM
conference on Electronic commerce, 1999.

• B. Sarwar, G. Karypsis, J. Konstan, & J. Riedl, Item-Based Collaborative Filtering Recommendation


Algorithms n WWW10, 2001.
THANK YOU

You might also like