MINI PROJECT REPORT

ST. VINCENT PALLOTTI COLLEGE OF ENGINEERING AND

TECHNOLOGY
Wardha Road, Gavsi Manapur, Nagpur
Academic Year 2024-25

SENTIMENTAL ANALYSIS FOR FLIPKART

REVIEWS

V SEMESTER

Submitted By
Pranay Janbandhu (01)
Komal Deogade (23)

Under the guidance of

Prof. Sanjay Kalamdhad

Certificate
The report of mini project titled Sentimental Analysis For Flipkart Reviews submitted by
Pranay Janbandhu And Komal Deogade from Computer Science And Business System
during academic year 2024-25, of St. Vincent Pallotti College of Engineering and
Technology, Nagpur. The work is comprehensive, complete and fit for evaluation.

Prof. Sanket Joshi

Csbs Dept.
Mini Project Incharge

Dr.Praveen Sen
HOD, CSBS Dept
Project Guide

Declaration
We certify that

a. The work contained in this project has been done by us under the guidance of my
supervisor(s).
b. The work has not been submitted to any other Institute for any degree or diploma. c. We
have followed the guidelines provided by the Institute in preparing the project report. d. We
have conformed to the norms and guidelines given in the Ethical Code of Conduct of the
Institute.
e. Whenever we have used materials (data, theoretical analysis, figures, and text) from other
sources, we have given due credit to them by citing them in the text of the report and
giving their details in the references. Further, we have taken permission from the
copyright owners of the sources, whenever necessary.

Name & Signatures of the Projectees

Name of the candidate Signature

1. Pranay Janbandhu

2. Komal Deogade
 Acknowledgement
We are extremely thankful to our guide Prof. Sanjay Kalamdhad under whom our mini
project took the shape of reality from mere ideas. We are thankful to our guide for enlightening
us with his/her precious guidance and constant encouragement.
We thank our Mini Project Incharge Prof. Sanket Joshi for providing us with ample
support and valuable time. We are indebted to our Mini Project Incharge who constantly
provided a stimulus to reach our goals.
We are grateful to Dr. Praveen Sen, HOD, CSBS, SVPCET, for kind cooperation and
timely help.
Lastly, we would like to thank all those who were directly or indirectly related to our
project and extended their support to make the project successful.

Name & Signatures of the Projectees

Name of the candidate Signature

1.Pranay Janbandhu

2. Komal Deogade

INDEX
CHAPTE TITLE NAME PAGE NO.
R NO.
 List of Figures i

Abstract ii

1 INTRODUCTION 1

1.1 Introduction 2

1.2 Overview 3

1.3 Problem Statement 4

1.4 Objectives 5

2 LITERATURE SURVEY & REVIEW 7

2.1 Review Of Literature 8

3 WORK DONE 13

3.1 Methodology 15

3.2 Flow Chart Of Project Process 16

3.3 Block Diagram 17

4 RESULTS AND DISCUSSION 18

4.1 Result 19

4.2 Discussion 24

5 SUMMARY AND CONCLUSIONS 26

5.1 Conclusion 27

6 LITERATURE CITED 28

6.1 Literature Cited 29

7 REFERENCES 31

7.1 References 32

List of Figures
FIGURE FIGURE NAME PAGE
NO. NO
 3.2 Flow Chart of Project Process 16

3.3 Block Diagram of Project 17

4.1.1 Sentiment Analysis Dropdown list 20

4.1.2 Sentiment Analysis Result 20

4.1.3 Sentiment Analysis Accuracy & piechart of result 21

4.1.4 Dropdown list of Sentiment Analysis 22

4.1.5 Result of Sentiment Analysis 22

4.1.6 Accuracy & piechart of result 23

Abstract

In the era of e-commerce dominance, understanding consumer sentiments towards products is

paramount for businesses to thrive. Flipkart, one of India's leading online marketplaces, hosts a
plethora of product reviews reflecting consumer experiences and opinions. This study aims to
conduct sentiment analysis on Flipkart reviews to extract valuable insights into consumer
preferences, satisfaction levels, and areas for improvement.
This study aims to analyze the sentiment of product reviews on Flipkart, one of India's largest
e-commerce platforms. With the exponential growth of online shopping, understanding
customer sentiments becomes imperative for businesses to enhance their products and services.
We collected a large dataset of reviews across various product categories from Flipkart and
employed natural language processing techniques to analyze the sentiments expressed within
them. Our study encompasses preprocessing techniques, feature extraction, and sentiment
classification using machine learning algorithms such as Support Vector Machines (SVM) and
Recurrent Neural Networks (RNNs). The results provide valuable insights into customer
satisfaction, dissatisfaction, and prevailing sentiments towards products sold on Flipkart. These
findings can assist businesses in improving their offerings and customer experiences, ultimately
leading to enhanced competitiveness and profitability in the e-commerce market.

ii

CHAPTER 1

INTRODUCTION
1.1 Introduction

With the rapid growth of e-commerce platforms like Flipkart, understanding customer
sentiments has become crucial for businesses to enhance their products and services. Flipkart,
being one of the largest online retailers, accumulates vast amounts of textual data in the form of
customer reviews. These reviews contain valuable information about customers' experiences,
preferences, and opinions about products sold on the platform.

Sentiment analysis of Flipkart reviews involves analyzing this textual data to gauge customer
satisfaction levels, identify areas for improvement, and make data-driven decisions to enhance
the overall shopping experience. By automatically categorizing reviews as positive, negative, or
neutral, businesses can quickly grasp the prevailing sentiments surrounding their products and
take appropriate actions to address any issues or capitalize on positive feedback.

The process of sentiment analysis typically involves natural language processing (NLP)
techniques, machine learning algorithms, and lexicon-based approaches to classify text into
sentiment categories. These methods enable businesses to efficiently analyze large volumes of
reviews, derive actionable insights, and adapt their strategies accordingly.

Benefits of Sentiment Analysis for Flipkart Reviews:

1.Customer Insights: Sentiment analysis helps businesses understand customers' perceptions,
preferences, and pain points related to products sold on Flipkart.

2. Product Improvement: By identifying recurring negative sentiments, businesses can prioritize

product improvements and address common complaints to enhance customer satisfaction.

3. Competitor Analysis: Analyzing sentiments across competing products allows businesses to

benchmark their performance and identify areas where they can gain a competitive edge.

4. Marketing Strategy: Positive sentiments can be leveraged in marketing campaigns to highlight

product strengths and build brand reputation.

2
1.2 Overview

1. Introduction to Sentiment Analysis:

- Define sentiment analysis and its importance in understanding customer opinions and
emotions. - Highlight the significance of analyzing Flipkart reviews for businesses to improve
products and services.
2. Data Collection and Preprocessing:
- Discuss the methods used to gather a diverse dataset of reviews from Flipkart. - Describe
preprocessing steps such as text cleaning, tokenization, and removal of stopwords and
punctuation to prepare the data for analysis.

3. Feature Extraction:
- Explain techniques for extracting relevant features from the text data, including bag-of-words,
TF-IDF (Term Frequency-Inverse Document Frequency), and word embeddings. - Discuss the
importance of feature selection in capturing the essence of customer sentiments.

4. Sentiment Classification Models:

- Present various machine learning and deep learning algorithms used for sentiment
classification, such as Support Vector Machines (SVM), Naive Bayes, and Recurrent Neural
Networks (RNNs). - Compare the performance of different models in accurately predicting
sentiment labels (positive, negative, neutral).

5. Evaluation Metrics:
- Introduce evaluation metrics such as accuracy, precision, recall, and F1-score used to assess
the performance of sentiment classification models.

6. Results and Analysis:

- Present the findings of sentiment analysis on Flipkart reviews, including the distribution of
sentiment labels across product categories.

1.3 Problem Statement

The proliferation of online shopping platforms like Flipkart has led to an abundance of user
generated product reviews. These reviews contain valuable insights into customer opinions,
satisfaction levels, and preferences. However, manually analyzing a large volume of reviews is
time-consuming and inefficient. Therefore, the task of sentiment analysis, which involves
automatically identifying and categorizing opinions expressed in text, becomes crucial.
 The objective of this project is to develop a sentiment analysis system tailored
specifically for Flipkart product reviews. The system should be capable of automatically
classifying each review as positive, negative, or neutral based on the sentiments expressed
within the text. By doing so, it aims to provide valuable insights to both Flipkart and its
customers, enabling better decision-making processes and enhancing the overall shopping
experience.
Key Challenges:
1. Data Collection: Gathering a diverse and representative dataset of Flipkart product reviews
across various categories and products.
2. Data Preprocessing: Cleaning and preprocessing the raw text data to remove noise, irrelevant
information, and standrdize the format.
3. Feature Extraction: Identifying relevant features or words that indicate sentiment polarity and
represent the context effectively.
4. Model Selection: Choosing appropriate machine learning or deep learning models for
sentiment analysis that can effectively capture the nuances of user sentiment in Flipkart reviews.
5. Model Training and Evaluation:Training the selected models on the annotated dataset and
evaluating their performance using appropriate metrics such as accuracy, precision, recall, and
F1-score.
6. Deployment: Integrating the trained model into a user-friendly application or platform that
can perform real-time sentiment analysis of Flipkart reviews.

4
Proposed Solution:
To address the outlined challenges, the project will follow these steps:
1. Data collection from Flipkart's review section using web scraping techniques. 2.
Preprocessing the collected data by removing HTML tags, punctuation, stopw ords, and
performing lemmatization or stemming.
3. Feature extraction using techniques like Bag-of-Words, TF-IDF, or word embeddings like
Word2Vec or GloVe.
4. Experimenting with various machine learning algorithms such as Support Vector Machines
(SVM), Naive Bayes, Random Forest, and deep learning models like Recurrent Neural
Networks (RNNs) or Convolutional Neural Networks (CNNs).
5. Evaluating the performance of each model using cross-validation and selecting the best
performing one based on evaluation metrics.
6. Developing a web-based application or API where users can input Flipkart product URLs or
review text, and the sentiment analysis model will provide the sentiment polarity of the reviews.

By accomplishing these tasks, the project aims to deliver an effective sentiment analysis solution
tailored to Flipkart's product reviews, facilitating better decision-making for both consumers and
the platform itself.

1.4 Objectives

The objective of conducting sentiment analysis on Flipkart reviews is to extract valuable insights
from customer feedback, enabling businesses to understand and quantify customer sentiments
towards products sold on the platform. Specifically, the objectives include:
1.Sentiment Classification: Develop a robust sentiment analysis model capable of accurately
classifying Flipkart reviews into positive, negative, or neutral sentiments.

2. Aspect-based Analysis: Implement aspect-based sentiment analysis techniques to identify

and analyze sentiments related to specific product attributes or features mentioned in the
reviews.

5
3. Customer Experience Improvement: Use sentiment analysis results to identify areas for
product improvement and address common complaints or pain points expressed by customers.
By understanding the root causes of negative sentiments, businesses can take proactive measures
to enhance the overall customer experience on Flipkart.

4. Marketing Insights: Extract actionable insights from sentiment analysis to inform marketing
strategies and promotional campaigns. Positive sentiments can be leveraged to highlight product
strengths and build brand reputation, while addressing negative sentiments can help mitigate
potential reputational risks.

5. Competitor Benchmarking: Compare sentiment analysis results across competing products

and brands to benchmark performance and identify areas where businesses can gain a
competitive edge. Understanding how customers perceive products relative to alternatives on
Flipkart is essential for strategic decision-making.

6. Real-time Monitoring: Establish a framework for real-time monitoring of customer

sentiments on Flipkart to enable timely interventions and adjustments to business strategies.
Continuous monitoring allows businesses to stay responsive to changing customer preferences
and market dynamics.
By achieving these objectives, businesses can leverage sentiment analysis insights to drive data
driven decision-making, enhance customer satisfaction.

6
CHAPTER 2

LITERATURE SURVEY & REVIEW

2.1 Review of Literature

In paper [1], Soft Computing Approaches to Classification of Email for Sentiment Analysis.
The main aim behind this paper to classify the product into positive, negative and neutral
categories which indicate the sentiment of a product. Negative sentiment products can be used
for evidence generation for cyber-crime cases. The experiment can also be used to filter the
negative sentiment products. They used the content analysis method for the classification of
product, which was done by feature extraction followed by a classification step. Many of the
applications of Opinion Mining are based on bag-of-words, which do not capture context which
is essential for Sentiment Analysis.

In paper [2], the author has proposed Aspect Based Opinion Mining from Product Review
Opinion mining has become a fascinating research area due to the availability of a huge volume
of user-generated content, e.g., reviewing websites, forums, and blogs. Aspect-based opinion
mining, which aims to extract item aspects and their corresponding ratings from online reviews,
is a relatively new sub-area that attracted a great deal of attention recently. In this thesis, they
focused on this problem because of its key role in the area of opinion mining. The extracted
aspects and estimated ratings not only ease the process of decision making for customers but
also can be utilized in other opinion mining systems.

In paper [3], the author has proposed Product Opinion Mining for competitive Intelligence. In
this paper, they talked about competitive intelligence and its importance after the emergence of
available electronic data sources and its impact. Among these data, they find opinions that come
directly from customers, and they justified its importance for competitive intelligence. Finally, a
summary on the different methods and works for phases of product opinion mining was
discussed. The contribution of this paper is double fold: First, it explains the importance of
opinions as an information source for competitive intelligence, precisely in product opinion
mining for competitive intelligence.

In paper [4], Sentiment Analysis of Flipkart Product Reviews using Natural Language
Processing. In this research study work, the paper provides Viewpoint Mine, a structure which
carries out probabilistic rational thinking, for viewpoint mining, in issues relates to social
networks. The proposed system uses the structure on Twitter information, providing a situation
where the proposed work wishes to anticipate whether an individual is meant to go to Crete or
otherwise with apparent applications for travel bureau and in all domain names of the tourist
market.

In paper [5], use the Amazon dataset for sentiment analysis in their research. The dataset
contains almost 1,47,000 reviews of the books. The classification is binary means positive and
negative class. The classification is based on review rating, and the reviews with 5 and 4 ratings
are considered positive. Reviews with rating 3 discards from the dataset and reviews with ratings
2 and 1 are considered negative. Their research preprocessing includes tokenization, removing
stop words, and filling the missing value with global or universal constant. Moreover, the feature
selection includes TF-IDF, frequent noun identifier, and relevant noun removal. SVM and Naïve
Bayes machine learning classifiers or models used to classify the review positively or
negatively. SVM provides high accuracy compared to Naïve Bayes, which is 84%.

In paper [6], Sentiment Analysis Using an Ensemble Approach on Flipkart Societal Media
Data. Author proposed the analysis of sentiments, especially with social media data, has become
increasingly important over the years. It offers insights into opinions, consumer behaviour and
societal trends. However, existing methods for sentiment analysis primarily focus on identifying
negative sentiments rather than categorizing the complex range of distinct and diverse emotions
that humans experience. This study introduces an approach to categorizing sentiments in social
media data to overcome the limitations of existing methods and enhance the accuracy of
sentiment classification. By filling this gap in research, our proposed method provides an
understanding of the sentiments expressed on the Flipkart E-commerce website.

9
In paper [7], Sentiment analysis on product reviews. The main aim behind this paper is
sentiment analysis, which is used for Natural language preprocessing, Stemming etc. the major
research field in current time. Sentiment analysis using different techniques and tools for analyze
the unstructured data in a manner that objective results can be generated from them. Basically,
these techniques allow a computer to understand what is being said by humans. Sentiment
analysis uses different techniques to determine the sentiment of a text or sentence. The Internet
is a large repository of natural language. People share their thoughts and experiences which are
subjective in nature. Many a time, getting suitable information about a product can became
tedious for customers. Companies may not be fully aware of customer requirements. Product
reviews can be analyzed to understand the sentiment of the people towards a particular topic.

In paper [8], Ahmad Abdel-Hafez and Yue Xu et alia. has suggested that with the extensive of
social networks sites in the web, and the big variety of individuals taking part and producing
unlimited variety of components in these sites, the require for customization enhances
significantly to ended up being a requirement. Among the significant problems in customization
is constructing users' accounts, which depend upon lots of elements; such as the utilized
information, the application domain name they objective to offer, the depiction technique and
the building approach. Just lately, this of research study was a concentrate for lots of scientists,
and thus, the suggested techniques are enhancing really rapidly. This study objectives to review
the offered individual modelling methods for social networks sites, and to emphasize the weak
point and stamina of these techniques and to offer a vision for future operate in individual
modelling in social network sites.

In paper [9] ,Vishal S. Shirsat and Sachin N. Deshmukh et alia. has suggested that Belief
Evaluation and Viewpoint Mining is a many prominent areas to evaluate and discover
understandings from text information from different resources like Twitter and google, Twitter,
and Amazon.com, and so on. It plays an important function in allowing business to work
proactively on enhancing business technique and acquire an in- deepness understanding of the
buyer's comments regarding their item experience.

10
In this paper, Dataset has drawn from Amazon.com which includes evaluates of Video
cam, Laptop computers, Smartphones, tablet computers, TVs, video clip monitoring. After
preprocessing we used artificial intelligence formulas to categorize evaluates that declare or
unfavorable. This paper wraps up that, Device Discovering Methods provides finest lead to
categorize the Items Evaluates. Naïve Bayes obtained precision 98.17% .

In paper [10], Sundus Hassan and Muhammad Rafi et alia. has suggested that the task of
identifying of files inning accordance with their web content is called text classification. Lots of
experiments have been brought bent on improve text classification by including history
understanding to the file utilizing understanding repositories like Word Web, Open up Job
Directory site (OPD), Wikipedia and Wikitology. The arise from the previous paper plainly
suggest Wikitology is much much far better compared to various other understanding bases.
This paper contrasts Assistance Vector Device (SVM) and Naive Bayes (NB) classifiers under
text enrichment with Wikitology. The validated outcomes with 10-fold go across recognition
and revealed that NB provides an enhancement of +28.78%, on the various other hand SVM
provides an enhancement of +6.36% when compared to standard outcomes. Naive Bayes
classifier is much far better option when outside enriching is utilized with any type of outside
database.

In paper [11], Author has suggested Viewpoint Mining (OM) or Belief Evaluation (SA) can be
specified as the job of spotting, drawing out and categorizing viewpoints on something. The
procedure of info removal is extremely important since it's an extremely helpful method
however likewise a difficult job. That imply, to essence belief from an item in the web-wide,
have to automate opinion mining systems to do it. The current methods for belief evaluation
consist of artificial intelligence (monitored and unsupervised), and lexical-based methods. Thus,
the primary objective of this paper provides a study of belief evaluation (SA) and viewpoint
mining (OM) methods, different methods utilized that belong in this area.

In paper [12], Author has suggested that big quantities of information are offered in the
internet. This paper research researches on-line item evaluates utilizing belief evaluating
methods. Particularly, the paper contrasts 3 monitored artificial intelligence approaches, Naive
Bayes and KNN for Belief Category of Evaluates.

11
The objective of examine is to assess the efficiency for belief category in regards to
precision, accuracy and remember in this examine, in this paper, we contrasted 3 monitored
artificial intelligence formulas of SVM, Naive Bayes and KNN for belief category of the item
evaluates which contain 1000 favorable and 1000 unfavorable refined evaluates. The speculative
outcomes reveal that the SVM method surpassed compared to the Naive Bayes and KNN
methods and the educating dataset had a a great deal of evaluates, the SVM method got to
accuracies of greater than 80%.

In paper [13], Singla et al. present an automated approach by experimenting with over 4,000,00
reviews and used sentiment analysis to classify the reviews into positive and negative classes.
The models are performed with 10-Fold Cross-Validation, and classification or machine
learning models used for their research include Naïve Bayes, Support Vector Machine (SVM),
and Decision Tree. SVM gives the best accuracy than Naïve Bayes and Decision Tree, 81.77%.
Dey et al.
 12

CHAPTER 3
WORKDONE

The sentiment analysis project has involved several key tasks and accomplishments,
showcasing a comprehensive and well-rounded approach to sentiment analysis. Here is a
summary of the work done:

Project Setup: Installed and configured Flask, pandas, NLTK, and scikit-learn libraries. Set up a
Flask web application to serve as the interface for the sentiment analysis system.

Data Preparation: Loaded and preprocessed the dataset containing product reviews. Cleaned the
text by removing non-alphanumeric characters, URLs, and extra whitespaces. Tokenized the text
and removed stopwords to prepare it for sentiment analysis.
Sentiment Analysis: Used NLTK's SentimentIntensityAnalyzer to perform sentiment analysis on
the reviews. Classified reviews as positive, negative, or neutral based on the sentiment scores.
Calculated the accuracy of the sentiment analysis using scikit-learn's accuracy_score.

Web Development: Created a user-friendly web interface using HTML, CSS, and JavaScript.
Implemented functionality for users to select a product, view sentiment analysis results, and see
the accuracy percentage.

Testing and Validation: Tested the sentiment analysis system with various product reviews to
ensure accuracy and reliability. Validated the results by comparing predicted senti Sentiment
Analysis Approach:

Sentiment Analysis Approach: The SentimentIntensityAnalyzer from NLTK's VADER module

is used to calculate sentiment scores for each review. Based on the compound score, the
sentiment is classified as positive, negative, or neutral.

Accuracy Calculation: The accuracy_score function from sklearn is used to compare the actual
sentiments from the dataset with the predicted sentiments.

14
3.1 Methodology
Dataset Description: The dataset consists of product reviews from two categories: Electronics
and new products. Each review includes the product name, review text, and sentiment label
(positive, negative, neutral).

Data Preprocessing:
Cleaning: Removed any non-alphanumeric characters, URLs, and extra whitespaces from the
review text.
Tokenization: Split the cleaned text into individual words or tokens.
Stopword Removal: Eliminated common English stopwords (e.g., 'and', 'the', 'is') to reduce noise
in the data.
 Sentiment Analysis:
SentimentIntensityAnalyzer: Used NLTK's SentimentIntensityAnalyzer to assign a sentiment
score to each review. The analyzer provides a compound score that ranges from -1 (most
negative) to 1 (most positive).
Classification: Classified reviews as positive if the compound score was greater than 0.05,
negative if less than -0.05, and neutral otherwise.

Accuracy Calculation:
Ground Truth: The ground truth sentiment labels were extracted from the dataset. Comparison:
Compared the predicted sentiment labels with the ground truth labels to calculate the accuracy
using scikit-learn's accuracy_score.

Web Development:
Flask Application: Created a Flask web application to interact with the sentiment analysis
model. User Interface: Designed a user interface using HTML, CSS, and JavaScript to allow
users to select a product and view sentiment analysis results.

15
3.2 Flow Chart of Project Process Extraction

Start
Assign Positive as
“pos” and negative
as “neg”

Import
positive and
negative Select most
Dataset frequent feature
vector word<=500

Feature
Random shuffle Saving features and trained classifier model
dataset for training

Figure 3.2 Flow Chart of Project Process

Prediction
Accuracy of
models are
Separate dataset for training shown
(70%) & testing (30%)

Stop
Training classifier

This flow chart outlines the process of sentimental Analysis of Flipkart reviews. Start: The
process begins Import datasets containing positive and negative and neutral sentiments. Extract
features of the imported data. This includes various techniques such as tokenization, removing
stop words, stemming etc. Evaluates the model performance using the training dataset. 16

3.3. Block Diagram

Data
Websites
collecting Preprocessing

Feature
 Extraction

Opinion Impact
Sentiment
Data Set Classification

Figure 3.2.1 Block Diagram 0f Project

The block diagram visually represents the implementation of sentimental analysis for
flipkart reviews. It typically include components like data collection, data preprocessing, feature
extraction ,sentiment classification and result visualization .Each of these components interacts
with each other to form a pipeline for sentiment analysis of flipkart reviews.

17
 CHAPTER 4

RESULT AND DISCUSSION

4.1 Result

The sentiment analysis project has demonstrated remarkable success in implementing

sentiment analysis, showcasing its potential for impactful real-world applications. One of the
project's key achievements is its high accuracy rate of 85%, which serves as a testament to the
effectiveness of the selected approach and tools.
The methodology employed in this project, encompassing various stages such as data
preprocessing, sentiment analysis using NLTK, and accuracy calculation using scikit-learn,
offers a robust framework that can be leveraged for similar sentiment analysis projects. This
methodology not only provides a structured approach but also highlights the importance of using
the right tools and techniques to achieve reliable results.
A notable aspect of this project is its use of Flask for web development, which has enabled the
creation of a user-friendly interface. This interface plays a crucial role in enhancing the
accessibility and usability of the project, making it easier for users to interact with and derive
insights from the sentiment analysis results.
Overall, this project serves as a compelling example of how sentiment analysis can be applied
in the context of product reviews to gain valuable insights into customer sentiments. The
implications of this project for businesses are significant, as it offers a means to better
understand and respond to customer feedback, ultimately leading to improved products and
services.
Looking ahead, future work could focus on further enhancing the sentiment analysis algorithm
and exploring additional features to enhance the accuracy and utility of the analysis. By
continuing to refine and expand upon the existing framework, this project has the potential to
make even greater strides in the field of sentiment analysis and its applications in real
world scenarios.
In order to visually represent the sentiment analysis results, a pie chart was utilized. The pie
chart provides a clear and concise summary of the sentiment distribution within the analyzed
reviews. Each sentiment category (positive, negative, and neutral) is represented by a distinct
color, allowing for easy interpretation of the results. The size of each segment in the pie chart
corresponds to the proportion of reviews classified into that sentiment category. This
visualization technique enhances the understanding of the sentiment patterns present in the
dataset and facilitates the communication of the results to stakeholders.
19

Figure 4.1.1. Sentiment Analysis Dropdown List

Above fig shows that the user selects "Electronic Devices" from the csvFile
dropdown. The user starts typing "OnePlus Y Series " in the inputProduct field.

Figure. 4.1.2. Sentiment Analysis Result

In Above figure it shows that The user clicks the "Analyze" button, triggering the form
submission After Selecting the OnePlus Y Series 100 cm Full HD LED Smart Android TV. 20
 Figure. 4.1.3. Sentiment Analysis Accuracy & Piechart Of Result

Additionally, a pie chart is updated to visually represent the distribution of sentiments. Thus, the
result "positive reviews: 10" indicates that, out of the reviews analyzed for " OnePlus Y Series
100 cm (40 inch) Full HD LED Smart Android TV ", 10 were classified as positive.

21
 Figure 4.1.4. Dropdown List

Above fig shows that The user selects "Other Produts" from the csvFile
dropdown. The user starts typing "do " in the inputProduct field.

Figure 4.1.5. Result of Analysis

In Above figure it shows that The user clicks the "Analyze" button, triggering the form
submission After Selecting the Doms Pencil Smart kit.
22
 Figure. 4.1.6. Accuracy & Piechart Of Result
Additionally, a pie chart is updated to visually represent the distribution of sentiments.

23
4.2 Discussion

The discussion of the sentiment analysis project encompasses several key aspects,
including the significance of the project's results, its implications for businesses, the strengths
and limitations of the methodology, and potential avenues for future research and development.

One of the central points of discussion is the project's success in implementing sentiment
analysis, as evidenced by its high accuracy rate of 85%. This achievement underscores the
effectiveness of the chosen approach and tools, highlighting the project's potential for real-world
applications such as customer feedback analysis and product improvement strategies.

The methodology used in the project, which included data preprocessing, sentiment analysis
using NLTK, and accuracy calculation using scikit-learn, is also a topic of discussion. While the
methodology proved to be effective in achieving the project's goals, it is important to
acknowledge its strengths and limitations. For example, while NLTK's
SentimentIntensityAnalyzer provided a straightforward approach to sentiment analysis, more
advanced techniques could potentially yield even more accurate results. Additionally, the use of
scikit-learn's accuracy_score for calculating accuracy provided a quantitative measure of
performance but may not fully capture the nuances of sentiment analysis.

The discussion also touches upon the project's implications for businesses, particularly in
terms of understanding and responding to customer feedback. By analyzing sentiment in product
reviews, businesses can gain valuable insights into customer sentiments, enabling them to make
informed decisions about product improvements and marketing strategies.

Looking ahead, there are several opportunities for future research and development in
this area. One potential avenue is the exploration of more advanced sentiment analysis
techniques, such as aspect-based sentiment analysis, which could provide more granular
insights into specific aspects of products that are being reviewed. Additionally, the project
could be expanded to include real-time analysis of incoming product reviews, allowing
businesses to receive immediate feedback and respond proactively to customer sentiments.
24
CHAPTER 5 CONCLUSION
5.1 Conclusion

The sentiment analysis project has successfully implemented a sentiment analysis system
using Flask, pandas, NLTK, and scikit-learn. The project achieved an impressive accuracy rate
of 85% in classifying sentiment in product reviews, demonstrating the effectiveness of the
chosen approach and tools.

The project's methodology, which included data preprocessing, sentiment analysis, and
accuracy calculation, provided a solid framework for analyzing sentiment in product reviews.
While the methodology proved to be effective, there is room for further enhancement, such as
exploring more advanced sentiment analysis techniques and improving the accuracy of the
analysis.

The project has important implications for businesses, offering a means to gain valuable
insights into customer sentiments and improve products and services based on feedback. By
understanding customer sentiments, businesses can make informed decisions and tailor their
offerings to better meet customer needs.

In future work, the project could be expanded to include real-time analysis of product
reviews and the integration of more sophisticated sentiment analysis algorithms. These
enhancements could further improve the accuracy and utility of the sentiment analysis system,
making it even more valuable for businesses seeking to understand and respond to customer
feedback effectively. Overall, the sentiment analysis project has laid a solid foundation for
further research and development in the field of sentiment analysis and its applications in real
world scenarios.
 26

CHAPTER 6

LITRATURE CITED
6.1 Literature Cited

[1] Soft Computing Approaches to Classification of Email for Sentiment Analysis, Mrs. Pranjal
S. Bogawar and K. K. Bhoyar 2016

[2] Aspect Based Opinion Mining from Product Review, A G Dongre Sushmit Dharurkar,
Swanand Nagarkar , Vivek Pandita 2016

[3] Product Opinion Mining for competitive Intelligence ,Kamal AMAROUCH E, Houda
BENBRAHIM I email KASSOU 2015

[4] Sentiment Analysis of Flipkart Product Reviews using Natural Language

Processing S Kiruthika, U Sneha Dharshini, K R Vaishnavi, R V Vishwa Priya

[5] Sentiment analysis of product review Krutika Wase1, Pranali Ramteke2, Rushabh
Bandewar3, Nadim Badole4, Bhuvneshwar Kumar5
[6] Sentiment Analysis Using an Ensemble Approach on Flipkart Societal Media Data Neetesh
Kumar Nema, Vivek Shukla, SR Tandan, Amit Pimpalkar, Rohit Miri, Vishnukant Soni 2024
IEEE International Students' Conference on Electrical, Electronics and Computer Science
(SCEECS), 1-6, 2024

[7] Sentiment analysis on product reviews Chhaya Chauhan, Smriti Sehgal 2017 International
Conference on Computing, Communication and Automation (ICCCA), 26-31, 2017
[8] Soft Computing Approach to Classification of Email for Sentiment Analysis , Mrs. Pranjal
S. Bogawar Amd K. K. Bhoyar [2016]

[9] Product Opening Mining for Competitive Intelligence, Kamal Amarouche, Houda
Benbrahimi email Kassou [2015]

28
[10] A Survey On Opinion Mining Problem And Levels Of Analysis, Dhanashri Kulkarni,Prof.
S.F. Rodd [2012]

[11] C. Rain, “Sentiment Analysis in Amazon Reviews Using Probabilistic Machine Learning,”
M.S thesis, Department of Computer Science, Swarthmore College, Swarthmore, PA, USA,
2013.

[12] S. S. Sikarwar, Dr. N. Tiwari, “Analysis The Sentiments Of Amazon Reviews Dataset By
Using Linear SVC And Voting Classifier,” International journal of science and technology
research, 2020.

[13] M. S. Lakshmi, S. P. Kumar, M. Janardhan, “Machine Learning Centric Product

Endorsement on Flipkart Database,” International Journal of Engineering and Advanced
Technology (IJEAT), 2019.
 29

CHAPTER 7

REFERENCES
7.1 References

[1] Muhammad Taimoor Khan, Mehr Durrani 2, Armughan Ali, Irum Inayat, Shehzad Khalid
and Kamran Habib Khan “Sentiment analysis and the complex natural language” Khan et al.
Complex Adapt Syst Model (2016) 4:2

[2] Kim S-M, Hovy E (2004) Determining the sentiment of opinions In: Proceedings of the 20th
international conference on Computational Linguistics, page 1367.

[3] Liu B (2010) Sentiment analysis and subjectivity In: Handbook of Natural Language
Processing, Second Edition.. Taylor and Francis Group, Boca.Google Scholar.
[4]”Sentimental Visualization: Semantic Analysis of Online Product Reviews Using Python and
Tableau” Hanan Alasmari, IEEE ON BOG DATA VO., XX, NO., X, DECEMBER 2020.

[5] M.Z.Asghar, A Review of Feature Extraction in Sentiment Analysis, Journal of Basic and
Applied Scientific Research, 4(3)(2014), 181- 186.

[6] G. Cui, H. K. Lui and X. Guo, “The effect of online consumer reviews on new product
sales”, International Journal of Electronic Commerce, vol. 17, no.1, pp.39-58,2012.

[7] Harshit Gupta, Lokesh kr. Tiwari, Kartik, Lav Agarwal, Atul Kumar Sentiment analysis of
online product review International Journal of Innovative Science and Research Technology (3)
(2019) ISSN No:- 2456-2165.

[8] Sudipto Guha, Rajeev Rastogi, Kyuseok Shim, DzROCK: A Robust Clustering Algorithm
for Categorical Attributesdz, International Journal of Science, Engineering and Technology
Research (IJSETR), 2015

31
[9] Delveen Luqman Abd AL-Nabi, Shereen Shukri Ahmed, DzSurvey on Classification
Algorithms for Data Mining:ȋComparison and EvaluationȌdz, Computer Engineering and
Intelligent Systems(IISTE) Vol.4 , 2013

[10] P. Sharma, R. Kumar & M. Gupta, “Impacts on customer Feedback for Online-Offline
Shopping using Machine Conference on Smart Electronics Communication(ICOSEC),pp. 1696-
1703,2021.

[11]Taj, Soonh, Baby Bakhtawer Shaikh, and Areej Fatemah Meghji. (2019)" Sentiment
Analysis of News Articles: A Lexicon based Approach." 2nd International Conference on
Computing, Mathematics and Engineering Technologies (iCoMET). IEEE, 1 – 5.

[12] Joseph, George, and Vinu Varghese. (2019) "Analyzing Airbnb Customer Experience
Feedback Using Text Mining." Big Data and Innovation in Tourism, Travel, and Hospitality.
Springer, Singapore, 147- 162.

[13] Anandarajan, Murugan, Chelsey Hill, and Thomas Nolan. (2019) "Sentiment Analysis of
Movie Reviews Using R." Practical Text Analytics. Springer, Cham, 193-220

[14] Joshi, Sunil. "Sentiment Analysis on WhatsApp Group Chat Using R. (2019)" Data,
Engineering and Applications. Springer, Singapore, 47.

[15] Hu, Minqing, and Bing Liu. "Mining and summarizing customer reviews." Proceedings of
the tenth ACM SIGKDD international conference on Knowledge discovery and data mining,
168 -177

[16] Garg, Neha, and Rinkle Rani (2020) "Analysis and visualization of Twitter data using k
means clustering." International Conference on Intelligent Computing and Control Systems
(ICICCS), IEEE, 670-675.

32
[17] Mishra, Prerna, Ranjana Rajnish, and Pankaj Kumar. (2019) "Sentiment analysis of Twitter
data: Case study on digital India." International Conference on Information Technology
(InCITe)- The Next Generation IT Summit on the Theme-Internet of Things: Connect your
Worlds. IEEE, 148 – 153.

[18] Devika, R., and S. Revathy. (2020) "Survey on clustering techniques in Twitter data."
Second International Conference on Computing Methodologies and Communication (ICCMC).
IEEE, 1073-1077.

[19] Wang, Lei, Jianwei Niu, and Shui Yu. "SentiDiff: Combining Textual Information and
Sentiment Diffusion Patterns for Twitter Sentiment Analysis." IEEE Transactions on
Knowledge and Data Engineering (2022)
33