DETECTING FAKE ACCOUNTS ON

SOCIAL MEDIA INSTAGRAM

Submitted in partial fulfillment of the requirements for the award of

Bachelor of Engineering degree in Computer Science and Engineering

By

BHAVYA K ( Reg.No - 39110454 )

NIKHITHA K ( Reg.No – 39110443 )

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

SCHOOL OF COMPUTING

SATHYABAMA
INSTITUTE OF SCIENCE AND
TECHNOLOGY (DEEMED TO BE
UNIVERSITY)
Accredited with Grade “A” by NAAC | 12B Status by UGC | Approved by
AICTE
JEPPIAAR NAGAR, RAJIV GANDHI SALAI,
CHENNAI - 600119

APRIL - 2023
 DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
BONAFIDE CERTIFICATE

This is to certify that this Project Report is the bonafide work of Bhavya k (Reg.No -
39110454) and Nikhitha k (Reg.No - 39110443) who carried out the Project Phase-2
entitled “DETECTING FAKE ACCOUNTS ON SOCIAL MEDIA -
INSTAGRAM” under my supervision from January 2023 to April 2023.

Internal Guide

Dr. D. USHA NANDINI M.E., Ph.D.

Head of the Department

Dr. L. LAKSHMANAN, M.E., Ph.D.

Submitted for Viva voce Examination held on 20.04.2023

Internal Examiner External Examiner

 DECLARATION

I, Bhavya k(Reg.No- 39110454), hereby declare that the Project Phase-2 Report
entitled “DETECTING FAKE ACCOUNTS ON SOCIAL MEDIA - INSTAGRAM”
done by me under the guidance of Dr. D. Usha Nandini M.E., Ph.D., is
submitted in partial fulfillment of the requirements for the award of Bachelor of
Engineering degree in Computer Science and Engineering.

DATE: 20.04.2023
PLACE: Chennai SIGNATURE OF THE CANDIDATE
 ACKNOWLEDGEMENT

I am pleased to acknowledge my sincere thanks to Board of Management of

SATHYABAMA for their kind encouragement in doing this project and for
completing it successfully. I am grateful to them.

I convey my thanks to Dr.T.Sasikala M.E., Ph.D., Dean, School of Computing,

Dr.L.Lakshmanan M.E., Ph.D., Head of the Department of Computer Science
and Engineering for providing me necessary support and details at the right time
during the progressive reviews.

I would like to express my sincere and deep sense of gratitude to my Project

Guide Dr.D.Usha Nandini M.E., Ph.D., for her valuable guidance, suggestions
and constant encouragement paved way for the successful completion of my
phase-2 project work.

I wish to express my thanks to all Teaching and Non-teaching staff members of

the Department of Computer Science and Engineering who were helpful in
many ways for the completion of the project.
.
 ABSTRACT

As the growth of online social networks increases everyone is associated and

linked with social media. A massive amount of personal data is being attacked
and stolen by cyber attackers. These fraudulent profiles also spread negative
publicity, false news, and various malicious programs one of the best ways is to
detect these fake accounts, not only the best way but also it is the first step to
stopping the negative news and false rumours. There are many efficient ways and
techniques to detect these fake accounts. However, these techniques are based
on or depend on the account features. In this paper, main aim is to examine three
machine learning algorithms that are random forest, logistic regression, and
decision tree. Now, progressing to compare the level of efficiency in the three
mentioned methodologies and got the highest accuracy for random forest
classifier.

The Proliferation of fake accounts on social media platforms like instagram has
become significant challenge for both users and platform administrators. Fake
accounts are created for a variety of purposes, including spreading spam,
disseminating false information, and engaging in fraudulent activity. These
accounts can also negatively impact the user experience by generating fake
engagement, flooding comment sections with spam, and diluting the quality of
content on the platform. Therefore, it become crucial to develop effective
methods for detecting and removing fake accounts from instagram.
 TABLE OF CONTENTS

Chapter Page
TITLE
No. No.
ABSTRACT v

LIST OF FIGURES viii

LIST OF TABLES ix

1 INTRODUCTION 1
LITERATURE SURVEY
2 4
2.1 Inferences from Literature Studies Survey

2.2 Open problems in Existing System 7

3 REQUIREMENTS ANALYSIS
3.2 Feasibility /Risk Analysis of the Project 9
3.2.1 Feasibility Studies 9
3.2.2 Risk Analysis 10
3.3 Software Requirements Specification Document
3.3.1 Hardware requirements 11
3.3.2 Software requirements 11
3.4 System Usecase 12
4 DESCRIPTION OF PROPOSED SYSTEM
4.2 Methodology or process model 13
4.3 Architecture of the Proposed System 15
Description of Software for Implementation
4.3 15
and Testing plan of the Proposed
Model/System
4.3.1 System 16
4.3.2 User 16
4.4 Project Management Selected Plan 17
5 IMPLEMENTATION DETAILS
5.2 Development and Deployment Setup 18
5.1.1 PyCharm 18

5.1.2 Python 21
5.1.3 Flask Framework 23
5.1.4 Libraries Used 25
 5.3 Algorithms 30
5.4 Testing 35
5.4.1 Unit Testing 35
6 RESULTS AND DISCUSSION 37
7 CONCLUSION
7.2 Conclusion 42
7.3 Future work 42
7.4 Research Issues 43
7.5 Implementation Issues 44
REFERENCES 46
APPENDIX
A. SOURCE CODE 48
B. SCREEN SHOTS 54
C. RESEARCH PAPER 58
 LIST OF FIGURES

FIGURE FIGURE NAME Page No.

NO.
3.1 Usecase Diagram 12

4.1 System Workflow 14

4.2 System Architecture. 15

5.1 PyCharm Installation 20

5.2 Pycharm Setup 21

5.3 Python Installation 22

5.4 Types of nodes in decision tree 33

5.5 Example of decision tree 34

6.1 Home Page 37

6.2 About Page 38

6.3 View Page 38

6.4 Model Selection 39

6.5 Prediction 39

6.6 Logistic Regression accuracy 40

6.7 Random Forest accuracy 41

6.8 Decision Tree accuracy 41

 LIST OF TABLES

TABLE TABLENAME Page No.

NO.
4.1 Project plan 17

6.1 Accuracy 40
 CHAPTER 1

INTRODUCTION

In recent times, some of the most well-known social network sites like Instagram,
Facebook and Twitter became integral parts of daily life. Social network sites are
used by people for e-commerce, entertainment, information and idea sharing
keeping in contact with long-lost friends and finding new acquaintances. Users of
OSN can post pictures and videos, leave comments and also provide likes to
pictures that have been shared. Fraudulent accounts are one of the downsides of
Online Social Networks. We can check an account is fake or not by following
parameters, the number of followers, and posts, or the characteristics of the
posted information such as the number of likes, views, comments, and shares. By
knowing this, many fake accounts are formed and started purchasing comments
and increasing the number of followers to become popular. These are particularly
created to commit fraudulent acts such as disseminating misleading information
and spreading malware. While some are created to gain followers to become
popular, spam comments and likes. In social media, one individual creates many
profiles with distinct identities, email addresses, and phone numbers. The majority
of them are likely fake accounts. There are two sorts of fake accounts: duplicate
accounts and fraudulent accounts. Users create multiple accounts as secondary
accounts to promote their e-businesses. People and influencers establish
duplicate accounts to enhance their businesses and spread useful information.
This type of account is not harmful and does not violate social media terms and
conditions. Users create fake accounts mainly to spread negativity, false news,
hate speech, and impersonate others; these accounts might be considered
dangerous and called as false accounts.

This work primarily concentrated on detecting if an Instagram account is phony or

real. This detection is based on some parameters such as username, private or
public account, and number of followers, posts, follows and profile pictures. We
can detect a user's activity using this attributes. The user’s behaviour on
Instagram is determined by a collection of 12 features. Three machine learning
techniques are used to distinguish between genuine and fraudulent accounts.
For the period between 2014 and 2018 around 2.53 million U.S. dollars have been
spent on sponsoring political ads on Facebook by non-profits. The open nature of
OSNs and the massive amount of personal data for its subscribers have made
them vulnerable to Sybil attacks. In 2012, Facebook noticed an abuse on their
platform including publishing false news, hate speech, sensational and polarizing,
and some others. However, online Social Networks (OSNs) have also attracted the
interest of researchers for mining and analyzing their massive amount of data,
exploring and studying users’ behaviors as well as detecting their abnormal
activities. In researchers have made a study to predict, analyses and explain
customer’s loyalty towards a social media-based online brand community, by
identifying the most effective cognitive features that predict their customer’s
attitude. Facebook community continues to grow with more than 2.2 billion
monthly active users and 1.4 billion Daily active users, with an increase of 11% on
a year- over-year basis. In the second quarter of 2018 alone, Facebook reported
that its total revenue was $13.2 billion with $13.0 billion from ads only. Similarly, in
second quarter of 2018 Twitter has reported reaching about one billion of Twitter
subscribers, with 335 million monthly active users. In 2017 twitter reported a
steady revenue growth of 2.44 billion U.S. dollars, with 108 million U.S. dollars
lower profit compared to the previous year. In 2015 Facebook estimated that
nearly 14 million of its monthly active users are in fact undesirable, representing
malicious fake accounts that have been created in violation of the website’s terms
of service. Facebook, for the first time, shared a report in the first quarter of 2018
that shows their internal guidelines used to enforce community standards
covering their efforts between October 2017 to March 2018, this report illustrates
the amount of undesirable content that has been removed by Facebook, and it
covers six categories: graphic violence, adult nudity and sexual activity, terrorist
propaganda, hate speech, spam, and fake accounts.

In recent years many celebrities and businesses have created their accounts on
Instagram, they use Instagram to grow their business and fans. Furthermore,
many of them and other famous users use it as a platform for advertising. When
someone is boosting the number of followers over a hundred thousand or
millions,
it is no surprise to use that person’s account as a lucrative earner. Instagram has
widely used for sharing photos and videos and is profitable for celebrities,
businesses, and people with a considerable number of followers. In the
meantime, this high profit made this platform prone to be the potential place to be
used for malicious activities.
Such versatility and spread for the proliferation of abnormal accounts, which
behave in unusual ways. Most academic researchers have mostly focused on
spammers and accounts, which put their efforts into spreading advertising, spam,
malware and other suspicious activities. These malicious accounts are usually
using automatic programs to improve their performance, hide their real identity,
and look like real users. In past years, media have reported that accounts of
celebrities, politicians, and some popular business has indicated suspicious
inflation of followers. Fake Instagram accounts specially used to increase the
number of followers of a target account
 CHAPTER 2
LITERATURE SURVEY

2.1 INFERENCES FROM LITERATURE SURVEY

Detection of fake accounts is one of the major issues that should be solved as
soon as possible. Though various methods have been already making an impact,
these all methods are not completely accurate. Inspired by various research
papers on detection, in this whole literature review section, we are going to
discuss various applied methodologies and approaches.
Ersahin et.al [1], provided a categorization technique to find Twitter's Phony
accounts. And pre-processed our dataset by applying the Entropy Minimization
Discretization (EMD) method under supervision to numerical characteristics, and
then they examined the output of the Naive Bayes algorithm. By merely pre-
processing their dataset using the n discretization strategy on chosen
characteristics, they were able to improve the accuracy with Nave Bayes
from85.55% to.90.41% for a method for identifying bogus accounts on Twitter.
Only by applying the numerical information from provide a safe platform that can
forecast and spot fake news in social media networks.

Aditi Gupta et.al [2], used the Facebook Graph API, and collected Facebook user
feeds. The intricate privacy controls on Facebook made it extremely difficult to
gather data. On a minimal dataset that included their node, their acquaintances in
their social neighbourhood, as well as a collection of manually recognized spam
accounts, they performed the most widely used supervised machine learning
classification approaches. They assessed these classifiers’ performance to
identify the top classifiers that produced high detection rates and their capacity to
identify Phony Facebook accounts.

Yeh-Cheng Chen et.al [3], introduced a novel, efficient technique for detecting
bogus profiles using machine learning. Their method based on
account-by-account behavioural analysis evaluate whether an account is
genuinely fake or not accurately, as opposed to utilizing graph-based or manually
predicted that just
cover basic facts. On real-world data, their detection models work admirably, and
their findings indicate that the models are not over fitting.

Estee van der walt et.al, [4] employed artificial intelligence to identify Phony
identities. The author of this research clarified how to spot false profiles made by
both bots and humans. In this study, the fraudulent accounts made on social
networks by people and bots were compared. Finally, it can be said that while the
features employed to detect the bots were successful, they weren’t entirely
effective.
Lu Zhang et.al [5], suggested the PSGD, a partially supervised model, to identify
scammer groups. As a scammer group detector, a classifier is studied using PU-
Learning via PSGD. The suggested PSGD is effective and surpasses cutting-edge
spammer detection techniques, according to tests on a real-world data set from
Amazon.cn.Bucket online media, they believe it to be a significant and extremely
promising outcome.

Sarah Khaled et.al [6], investigated effective methods for spotting bots and Phony
accounts on the Twitter network. An innovative technique called SVM-NN is
recommended in this to offer an effective detection for such profiles. Even though
the recommended approach uses fewer features, it can still be categorized with
around ninety- eight percent accuracy. In comparison to the other two classifiers,
the newly suggested method performs better in terms of accuracy across all
feature sets. The accuracy of the correlation feature set's records is astounding.

Zulfikar Alom et.al [7], investigated the nature of spam users on the Twitter
platform to develop an existing spam-detecting mechanism. In this paper, it is
designed a new mechanism and a more robust set of features to detect
spammers on Twitter. In this Random Forest classifier is used which gives a better
result compared to other methodologies. From this, they can plan to build a more
effective model which classifies various types of spammers within different social
networks.
Naman Singh et.al [8], proposed that on online networking platforms, more bogus
accounts are being created for nefarious purposes. They create a forged human
profile in order to successfully detect, identify, and remove fake profiles. For bots,
ML models employed a variety of factors to determine how many people an
account has on each platform’s friend list as followers. It is not possible to
contrast between Phony profiles made by humans and cyborgs. By using a data
set with Phony profiles and classifying them as fake or real, they can contrast fake
and real accounts.

Shivangi Singhal et.al [9], formally introduced unveil Spot Fake a multi-modal
platform for identifying bogus news. Without considering any other subtasks, their
suggested approach detects bogus news. It makes use of article textual and
graphic components. Text characteristics were learned using language models
(such as BERT), and image features were learned using VGG-19 pre-trained on
Image Net dataset. Twitter and Weibo, two publicly accessible databases, are
used in all of the studies.
Sowmya P et.al [10], one serious issue involves generating duplicate profile users
using the data of existing users. A collection of principles that may differentiate
real or fake are used to detect Phony profiles. A detection method that can find
fraudulent or clone profiles on online networking platforms like Twitter has been
presented.

Zeinab Shahbazi et.al [11], proposed an integrated system for multiple block
chains and NLP functions to leverage ML models to find false news and better
anticipate bogus user accounts and posts. For this workflow, Reinforcement
learning methodology is applied. The operation of the decentralized block chain
architecture, which offers the framework of digital content authorization proof,
improved the security of this platform. The idea behind this approach is to provide
a safe platform that can forecast and spot fake news on social media networks.

Latha P et.al [12], stated different categorization techniques have been used in the
past to identify Phony online networking media accounts. However, they must
improve their ability to spot Phony accounts on these websites. To improve the
accuracy rate of identifying bogus accounts, they use ML technologies and NLP
in our study. The Random Forest tree classifying algorithm was chosen.
Michael Jonathan Ekosputra et.al [13], study's findings make it evident that ml
may be used to detect fake profiles; the models that have been examined in this
article include Logistic Regression, Bernoulli Naive Bayes, Random Forest, svm,
and ANN. Any changes or additions to features will have an impact on each
model's accuracy. According to the research, adding parameters to a model will
almost certainly make it more accurate than a model with no parameters or a
default form. Based on the results of the research's initial trial, the Random Forest
algorithm achieves the best outcome with a startling accuracy of 0.92.

Karishma Anklesaria et.al [14], used different classifiers to classify the profile as
fake or real using database of user accounts. This is accomplished by using a
methodical approach that includes cleaning, pre-processing, feature selection,
and model training. After that, all the employed algorithms—Random Forest,
AdaBoost, MLP, SGD, and Artificial neural network—are compared on the basis of
different evaluation parameters. According to the study, Random Forest classifier
performs better.

Md Mahadi Hassan Sohan et.al [15], showed the importance of comments and
how they affect almost every aspect of social media data. People's perspectives
are significantly influenced by reviews. Hence, identifying fake reviews is an
exciting study area. A method for spotting fake reviews using machine learning
was disclosed in the study. It detects both review characteristics and reviewer
behaviour. The food review dataset was used to evaluate the proposed method.
The technique is semi-supervised, with several classifiers being used. In the fake
review detection procedure, the findings reveal that the Random Forest classifier
beats another classifier. Moreover, the findings imply that including in viewers'
behavioural traits raises the F-score by 55.5% and the overall accuracy to 97.7%.

2.2 OPEN PROBLEMS IN EXISTING SYSTEM

The factors used by the existing systems to detect the fake accounts are very
less. The prediction becomes accurate when the number of parameters used are
more efficient. In previously used algorithms, if some of the inputs are not
appropriate, the algorithm could not produce accurate results.
● Low Accuracy.
● High complexity.
● Highly inefficient.
● Requires skilled persons.
 CHAPTER 3
REQUIREMENT ANALYSIS

3.1 FEASIBILITY STUDIES/RISK ANALYSIS OF THE PROJECT

The feasibility of the project is analysed in this phase and business proposal is
put forth with a very general plan for the project and some cost estimates. During
system analysis the feasibility study of the proposed system is to be carried out.
This is to ensure that the proposed system is not a burden to the company. For
feasibility analysis, some understanding of the major requirements for the system
is essential.

3.1.1 Feasibility Studies

Four key considerations involved in the feasibility analysis are

♦ TECHNICAL FEASIBILITY
♦ FINANCIAL FEASIBILITY
♦ LEGAL FEASIBILITY
♦ OPERATIONAL FEASIBILITY
Technical Feasibility: The technical feasibility of the project involves assessing
the availability of suitable data, machine learning algorithms, and hardware
resources. It is important to ensure that the data used in the project is reliable and
representative of the population of interest. The machine learning algorithms used
should be appropriate for the task of fake account detection and should be able
to handle large amounts of data. The hardware resources should be sufficient to
handle the computational requirements of the machine learning algorithms.

Financial Feasibility: The financial feasibility of the project involves evaluating

the cost of acquiring data, hardware resources, and expertise required for the
project. The cost of acquiring data can be significant, particularly if it is collected
using methods such as surveys or experiments. The cost of hardware resources
such as servers and computing power should also be considered. The cost of
expertise required for the project, such as data scientists and machine learning
experts, should also be evaluated.
Legal feasibility: The legal feasibility of the project involves investigating the legal
implications of the project. The project should be compliant with relevant laws
and regulations, particularly those related to privacy and data protection. The
project should also adhere to ethical guidelines, particularly those related to the
collection and processing of personal data.

Operational Feasibility: The operational feasibility of the project involves

determining the availability of necessary resources, such as human expertise and
technological infrastructure. The project should have access to experts in data
collection, data analysis, and machine learning. The project should also have
access to suitable technological infrastructure, such as servers and computing
power, to handle the computational requirements of the machine learning
algorithms.

3.1.2 Risk Analysis

Data quality risks: The quality of the data used to train the machine learning
model can affect the accuracy of the model. The data used to train the model
should be representative of the population of interest and should be free from
bias. The data should also be of high quality, with minimal missing values and
errors.

Model accuracy risks: The accuracy of the model depends on the selection of
the right machine learning algorithms, feature engineering, and parameter tuning.
The machine learning algorithms used should be appropriate for the task of fake
account detection, and should be able to handle large amounts of data. Feature
engineering should be carefully performed to identify the most relevant features
for the task. Parameter tuning should be performed to optimize the performance
of the model.

Privacy and ethical risks: The collection and processing of personal data raises
ethical and privacy concerns. The project should adhere to ethical guidelines,
particularly those related to the collection and processing of personal data. The
project should also ensure that appropriate privacy policies are in place to protect
the privacy of individuals whose data is collected.
3.2 SOFTWARE REQUIREMENTS SPECIFICATION DOCUMENT

3.2.1 Hardware Requirements

Processor - I3/Intel Processor

RAM - 4GB(min)
Hard Disk - 160GB
Keyboard - Standard Windows
Keyboard Mouse - Two or Three Button Mouse
Monitor - SVGA

3.2.2 Software Requirements

Operating System : Windows 10

Server-side Script : Python
IDE : PyCharm
3.3 SYSTEM USECASE

Fig. 3.1 Usecase Diagram

 CHAPTER 4

DESCRIPTION OF PROPOSED SYSTEM

4.1 SELECTED METHODOLOGY OR PROCESS MODEL

An effective method to detect fake accounts on online social networks is

proposed. This work may be regarded as a valuable system since it aids in
lowering the restrictions brought on by conventional and other existing
methodologies. The goal of this project is to provide an efficient and dependable
system for precisely detecting fake. A potent algorithm in a Python-based
framework to develop this system is employed. The process is depicted in the
picture below.
The steps for executing the proposed work:

1. The required packages should be installed.

2. Establishing the issue association.

3. Create a framework-based UI.

4. Upload CSV file

5. View data

6. Select model

7. Output to predict.

This proposed work is completely based on Machine Learning approach which

detects whether the profile is fake or real one. To do this, three well-known
methodologies were compared. This proposed work has two phases, the first
phase is frontend – which creates a platform another is backend – a model is
created and the main part of the project exists.
The main part of the proposed work is the backend. Here, using flask framework
an application is created that detects whether an account is fake or not. In this
aspect, the data is collected and preprocessed. Then the data splitting takes
place. Now the data gets trained and need to build the model. After the model
building with the help of the model data gets tested. And finally the prediction is
done.
Firstly, download the dataset from Kaggle which is raw data. The size of the
dataset is 697 with 12 attributes such as profile picture, length of the username,
full name words, length of the full name, the description provided by the user
known as bio, any external URLs provided, private or public accounts, number of
posts posted, number of followers and number of accounts the user follows.

In the application user first uploads a CSV file of the dataset then the system
takes that file and undergoes preprocessing, which means Importing libraries
such as pandas, and NumPy. Import dataset, cleaning of null values, categorical
values will be turned into numerical values and removing unwanted columns. The
null values are replaced by using imputation technique. Then splits the dataset
into training and testing set. The target variable or Dependent variable undergoes
training and independent variables will go to testing. Next is model building.
Model buildings mean choosing a suitable algorithm and train it as per the
requirements. Different models like decision tree, Random Forest, and Logistic
Regression are considered, and train the models by importing those modules
from Sklearn. Then, train the models using the training dataset with fit( ) function.
After the model building, apply that model to make predictions and predict the
response for test dataset. Then evaluate the model’s performance by importing
different metrics from sklearn metrics. Random forest got the highest accuracy
with 90%. So, this is how the system will do preprocessing and User can view the
dataset.
Fig. 4.1 Flow chart
4.2 ARCHITECTURE / OVERALL DESIGN OF PROPOSED SYSTEM

Fig. 4.2 System Architecture

4.3 DESCRIPTION OF SOFTWARE FOR IMPLEMENTATION AND

TESTING PLAN OF THE PROPOSED MODEL/SYSTEM

PyCharm is an extremely popular Python IDE. An Integrated Development

Environment or IDE features a code editor and a compiler for writing and
compiling programs in one or many programming languages. The main reason
PyCharm for the creation of this IDE was for Python programming, and to operate
across multiple platforms like Windows, Linux, and macOS. The IDE comprises
code analysis tools, debugger, testing tools, and also version control options. It
also assists developers in building Python plugins with the help of various APIs
available. The IDE allows us to work with several databases directly without
getting it integrated with other tools. Although it is specially designed for Python,
HTML, CSS, and JavaScript files can also be created with this IDE. It also comes
with a beautiful user interface that can be customized according to the needs
using plugins.
4.3.1. System

1. Store Dataset:
● The System stores the dataset given by the user, where the dataset
is downloaded from the kaggle by user.
2. Model Training:
● The system takes the data from the user and fed that data to the
selected model.
4.3.2 User
1. Load Dataset
● The user can load the dataset he/she want to work on.
2. View Dataset
● The size of the dataset is 697 with 12 attributes such as profile
picture, length of the username, full name words, length of the full
name, the description provided by the user known as bio, any
external URLs provided, private or public accounts, number of posts
posted, number of followers and number of accounts the user
follows. The user can view the dataset
3. Select model:
● User can apply the model among random forest, decision tree,
logistic regression to the dataset for accuracy. where random forest
got highest accuracy of 91.8%.
4. Prediction
● Passing parameters to predict the output

4.4 PROJECT MANAGEMENT PLAN

Instagram is the easiest platform to spread rumors, to stole one's information etc.,
these all kind of malicious are carried using fake accounts. So should be aware of
the account he/she interacting with. With the help of this project anyone can
detect fake accounts on instagram. In order to make the detection easily available
we
also implemented an application and also carefully considered all the existing
systems

In the project, the whole process is done in the pycharm platform. Compared
three well known models and flask framework to create an application. Collected
the dataset from kaggle which contains different users instagram information.

TABLE OF PROJECT MANAGEMENT PLAN

MODULES TIME TAKEN

Introduction September
Literature survey October
Software design November - December
Software implementation January - Febrauary
Testing March

Table no. 4.1 project plan

 CHAPTER 5
IMPLEMENTATION DETAILS
5.1 DEVELOPMENT AND DEPLOYMENT SETUP

5.1.1 Pycharm
PyCharm is a popular integrated development environment (IDE) for Python
programming language. Developed by JetBrains, PyCharm is designed to provide
developers with a wide range of tools and features to help them write, debug, and
maintain Python code more efficiently and effectively. In this article, we will take a
more in-depth look at the key features and functionalities of PyCharm.

User Interface
PyCharm has a clean and intuitive user interface that is designed to be user-
friendly and customizable. The IDE supports a range of themes and color
schemes, which can be customized to suit individual preferences. PyCharm also
supports multiple windows and tabbed interfaces, which allows developers to
work with multiple files and projects simultaneously.
Code Completion and Suggestion
One of the most powerful features of PyCharm is its intelligent code completion
and suggestion engine. The IDE can suggest method names, class names, and
variable names based on the context of the code. PyCharm can also provide
suggestions for import statements and function arguments. This feature can help
developers write code faster and with fewer errors.
Debugging Tools
PyCharm comes with a powerful debugger that allows developers to step through
their code, set breakpoints, inspect variables in real-time. The debugger also
supports remote debugging, which can be useful for debugging code running on
a remote server. PyCharm also provides support for Django, Flask, and Pyramid
frameworks, which allows developers to debug web applications directly from the
IDE.
Code Analysis and Error Highlighting
PyCharm has a built-in code analysis engine that can identify potential errors in
the code and highlight them for the developer. The IDE can flag issues like
undefined variables, unused imports, and syntax errors, which can help catch
bugs early and improve code quality. PyCharm also provides support for code
inspections, which can help identify issues like code smells, design problems,
and potential performance issues.
Version Control Integration
PyCharm has built-in integration with popular version control systems like Git,
SVN, and Mercurial. This allows developers to manage their code changes, track
changes over time, and collaborate with others on their projects. PyCharm
provides support for features like code merging, conflict resolution, and commit
history visualization.
Refactoring Tools
PyCharm provides a range of tools to help developers refactor their code. These
tools include renaming variables, extracting code into functions, and moving code
between files. PyCharm also provides support for safe delete, which allows
developers to delete code and its references without introducing errors in the
codebase.
Code Templates and Generation
PyCharm has a library of code templates that can be used to quickly generate
common code patterns, such as loops, if statements, and function definitions.
The IDE can also generate boilerplate code for popular frameworks like Flask and
Django. PyCharm also provides support for code generation from UML diagrams
and database schemas.
Testing Tools
PyCharm has built-in support for unit testing, including test runners, code
coverage analysis, and debugging tools for tests. The IDE can also generate test
stubs and test code automatically, based on the code being tested. PyCharm
also provides support for running tests remotely, which can be useful for testing
code running on a remote server.
Integration with Other Tools
PyCharm can be integrated with other tools in the development workflow, such as
code linters, task managers, and build tools. PyCharm provides support for
popular Python linters like Flake8, Pylint, and Pyflakes. The IDE can also be
integrated with build tools like setuptools, distutils, and pip.
In addition to these features, PyCharm also provides support for a range of other
functionalities, such as code documentation, code formatting, code search, and
code snippets. PyCharm is available in both a free, open-source community
edition and a paid professional edition with additional features and support. It
runs on Windows, macOS, and Linux, and it can be used for a wide range of
Python projects, from small scripts to large web applications.

Steps Involved for installation:

You will have to follow the steps given below to install PyCharm on your system.
These steps show the installation procedure starting from downloading the
PyCharm package from its official website to creating a new project.

Step 1

Download the required package or executable from the official website of

PyCharm https://www.jetbrains.com/pycharm/download/#section=windowsHere
you will observe two versions of package for Windows as shown in the
screenshot given below −

Fig. 5.1 Pycharm Installation

Note that the professional package involves all the advanced features and comes
with free trial for few days and the user has to buy a licensed key for activation
beyond the trial period. Community package is for free and can be downloaded
and installed as and when required. It includes all the basic features needed for
installation.
Step 2

Download the community package (executable file) onto your system and mention

a destination folder as shown below −

Fig.5.2 PyCharm Setup

PyCharm is the most popular IDE used for Python scripting language. This chapter
will give you an introduction to PyCharm and explains its features.

PyCharm offers some of the best features to its users and developers in the
following aspects −

● Code completion and inspection

● Advanced debugging
● Support for web programming and frameworks such as Django and Flask
5.1.2 Python

Python is a high-level, interpreted programming language that was first released

in 1991 by Guido van Rossum. Since then, it has become one of the most popular
programming languages in the world, with a wide range of applications, for
python is a high-level, interpreted programming language that was first released
in 1991 by Guido van Rossum. Since then, it has become one of the most popular
programming languages in the world, with a wide range of applications, from web
development and data science to machine learning and artificial intelligence.

Installing Python

1. To download and install Python visit the official website of

Fig.5.3 Python Installation

Python https://www.python.org/downloads/ and choose your version.

2. Once the download is complete, run the exe for install Python. Now click on
Install Now.

3. You can see Python installing at this point.

4. When it finishes, you can see a screen that says the Setup was successful.
Now click on "Close".
Here are some key features of Python:

Readable and easy to learn: Python has a simple syntax and is easy to learn for
beginners. Its code is also highly readable and expressive, making it easy to
understand and maintain.
Interpreted language: Python is an interpreted language, meaning that code is
executed directly by the interpreter, without the need for compilation.
Cross-platform: Python code can run on a wide range of platforms, including
Windows, Mac, Linux, and mobile devices.
Dynamic typing: Python is a dynamically typed language, meaning that variable
types are inferred at runtime rather than being explicitly declared.
Strong standard library: Python has a large standard library that provides a wide
range of functionality out of the box, including modules for string processing,
regular expressions, network programming, and more.
Object-oriented: Python is an object-oriented language, meaning that everything
in Python is an object, including functions and modules.
High-level: Python is a high-level language, meaning that it provides abstractions
that make it easier to express complex concepts without worrying about low-level
details.
Versatile: Python can be used for a wide range of applications, including web
development, scientific computing, data analysis, artificial intelligence, and more.
Large community: Python has a large and active community of developers, with
a wealth of resources and libraries available.
Open source: Python is an open-source language, meaning that its source code
is freely available and can be modified and distributed by anyone.

5.1.3 Flask Framework

Flask is a web application framework for Python, designed to make building web
applications easy and fast. It is a lightweight framework that is easy to learn and
use, making it a popular choice for small to medium-sized web applications.
Flask is built on top of the Werkzeug WSGI toolkit and the Jinja2 template engine,
which provides a powerful set of tools for handling requests and responses, as
well as rendering HTML templates.
Installation:
Install libraries-flask in pycharm
● Windows users can install flask via pip command:
pip install flask

One of the main features of Flask is its simplicity. Flask's API is very simple and
easy to use, with minimal boilerplate code required to get started. Flask is also
highly modular, allowing developers to easily add or remove components as
needed.
Flask provides a set of built-in tools for handling routing, input validation, form
processing, and more. It also supports a wide range of plugins and extensions,
making it easy to add additional functionality to your web application.
Flask uses a concept called "views" to handle requests from the user. A view is a
Python function that is decorated with the @app.route decorator, which specifies
the URL route that the function will handle. Views can return a variety of
responses, including HTML templates, JSON data, and more.
Flask also provides built-in support for handling authentication and security, with
support for OAuth, token-based authentication, and more. It also supports a
variety of databases, including SQLite, MySQL, and PostgreSQL, making it easy
to integrate your web application with a database.
Another feature of Flask is its support for testing. Flask provides a built-in testing
framework that allows developers to write automated tests for their web
application, ensuring that it is functioning correctly and reliably.

Overall, Flask is a powerful and flexible web application framework that is well-
suited for small to medium-sized web applications. Its simplicity, modularity, and
flexibility make it easy to learn and use, while its built-in tools and support for
plugins andextensions make it easy to add additional functionality as needed. If
you're looking to build a web application with Python, Flask is definitely worth
considering.
5.1.4 Libraries Used

1. Matplotlib

Matplotlib is an open-source data visualization library for the Python programming

language. It provides a wide range of tools for creating high-quality charts,
graphs, and other visualizations.
The core component of Matplotlib is the pyplot module, which provides a
convenient interface for creating plots and charts. Some of the most commonly
used types of plots that can be created using Matplotlib include line plots, scatter
plots, bar charts, histograms, and heatmaps. Matplotlib provides a lot of flexibility
in terms of customization options for plots, such as changing the colors, line
styles, and marker styles of data points, adding annotations, adjusting the size
and position of the plot, and much more.
To use the Matplotlib library in your Python code, you can import it using the
following command:
import matplotlib.pyplot as plt
Matplotlib provides a lot of customization options to make your plots look great.
You can change the colors, line styles, and marker styles of data points, add
annotations, adjust the size and position of the plot, and much more.
These are just a few examples of what you can do with Matplotlib. The library
provides a wide range of tools for creating all kinds of visualizations, from simple
line plots to complex 3D plots.
For 2D displays of arrays, Matplotlib is a fantastic Python visualization library. A
multi-platform data visualization package called Matplotlib was created to deal
with the larger SciPy stack and is based on NumPy arrays. One of visualization's
biggest advantages is that it gives us visual access to vast volumes of data in
forms that are simple to understand. There are numerous plots in Matplotlib,
including line, bar, scatter, histogram, etc. Install the matplotlib package by
running the command python -mpip install -U matplotlib, and then import
matplotlib. pyplot as plt.

2. Scikit-learn
Scikit-learn (also known as sklearn) is an open-source machine learning library for
the Python programming language. It provides a wide range of tools for building
and applying machine learning models, including supervised and unsupervised
learning, dimensionality reduction, feature selection, and model selection and
evaluation.
Scikit-learn includes implementations of many popular machine learning
algorithms, such as linear regression, logistic regression, decision trees, random
forests, support vector machines (SVMs), k-nearest neighbors (KNN), and
clustering algorithms (k-means, hierarchical clustering, etc.). It also includes tools
for preprocessing data (e.g., scaling, normalization, encoding categorical
variables, etc.) and for evaluating the performance of machine learning models
(e.g., cross-validation, metrics for classification and regression, etc.).
Scikit-learn is designed to be easy to use and provides a consistent interface for
all the algorithms it implements. It is built on top of other popular scientific
computing libraries in Python, such as NumPy, SciPy, and matplotlib, and is
widely used in both academia and industry for a variety of machine learning tasks.
To use the scikit-learn library in your Python code, you can import it using the
following command:
import sklearn
This will allow you to use all the functions and tools provided by scikit-learn.
Scikit-learn is an open-source data analysis library and the Python ecosystem's
gold standard for Machine Learning (ML). The following are key concepts and
features: Algorithmic decision-making techniques, such as:
Classification is the process of identifying and categorizing data based on
patterns. Regression is the process of predicting or projecting data values using
the average mean of existing and planned data.
Clustering is the automatic classification of similar data into datasets.
Installation:
Install libraries – scikit-learn in pycharm
● Users can install scikit-learn via pip command:
pip install scikit-learn

3. Pandas
Pandas is an open-source library designed primarily for working quickly and
logically with relational or labelled data. It offers a range of data structures and
procedures for working with time - series data and quantitative information. The
NumPy library serves as the foundation for this library. Pandas is quick and offers
its users exceptional performance & productivity. Checking to see if pandas is
installed in the Python folder is the first step in using it. If not, use the pip
command to install it on our machine. Enter the command cmd0 in the search
window, and then use the cd command to find the location where the python-pip
file is installed. Locate it and enter the following command: pip install panda and
then import panda as pd. In our project, Panda’s library helps us work with the
csv format database we have acquired.
Pandas is an open-source data manipulation and analysis library for the Python
programming language. It provides data structures for efficiently storing and
manipulating large datasets, as well as tools for data cleaning, merging,
reshaping, and aggregation.
The two primary data structures provided by Pandas are Series and DataFrame. A
Series is a one-dimensional labeled array that can hold any data type (integers,
floating-point numbers, strings, Python objects, etc.). A DataFrame is a two-
dimensional labeled data structure with columns of potentially different types. It is
similar to a spreadsheet or a SQL table.
Pandas provides a wide range of functions for working with data, including
filtering, sorting, grouping, joining, and reshaping. It also includes powerful tools
for data visualization, time series analysis, and data input/output.
To use the Pandas library in your Python code, you can import it using the
following command:
import pandas as pd
This will allow you to use all the functions and data structures provided by Pandas.
Installation:
Install libraries-pandas in pycharm
● Users can install pandas via pip command:
pip install pandas
5. NumPy

NumPy is a general-purpose array-processing package. It provides a high-

performance multidimensional array object, and tools for working with these
arrays. It is the fundamental package for scientific computing with Python.

NumPy (short for Numerical Python) is an open-source Python library that is used
for scientific computing and numerical analysis. It provides powerful tools for
working with arrays and matrices of numerical data, as well as a wide range of
mathematical functions for performing complex calculations.

To use the NumPy library in your Python code, you can import it using the
following command:

import numpy as np

This will allow you to use all the functions and data structures provided by
NumPy.It is open-source software. It contains various features including these
important ones:

● A powerful N-dimensional array object

● Sophisticated (broadcasting) functions
● Tools for integrating C/C++ and Fortran code
● Useful linear algebra, Fourier transform, and random number capabilities
Besides its obvious scientific uses, NumPy can also be used as anefficient
multi-dimensional container of generic data. Arbitrary data-types can be defined
using Numpy which allows NumPy to seamlessly and speedily integrate with a
wide variety of databases.
NumPy’s main object is the homogeneous multidimensional array.

● It is a table of elements (usually numbers), all of the same type, indexed by a

tuple of positive integers.
● In NumPy dimensions are called axes. The number of axes is rank.

● NumPy’s array class is called ndarray. It is also known by the alias

array. Installation:
Install libraries numpy in pycharm
● Windows users can install NumPy via pip command:
pip install numpy
4. Seaborn
Seaborn is a Python library for data visualization that is built on top of Matplotlib.
It provides a high-level interface for creating beautiful and informative statistical
graphics. Seaborn is particularly useful for exploring and visualizing complex
datasets with multiple variables.
Seaborn provides a wide range of built-in functions for creating different types of
visualizations, including scatter plots, line plots, bar plots, heatmaps, and more. It
also includes a number of advanced features, such as support for categorical
data, automatic estimation and plotting of regression models, and support for
multiple plot grids.
One of the key features of Seaborn is its support for creating aesthetically
pleasing visualizations. It provides a number of built-in color palettes that are
designed to be easy on the eyes, as well as a range of customizable themes that
can be used to adjust the overall look and feel of the visualizations.
Seaborn also provides a number of tools for visualizing relationships between
variables in datasets, including correlation matrices and pair plots. These tools
make it easy to identify patterns and relationships in data, and can help to guide
data analysis and modeling efforts.

Here are some of the specific plot types and functions that Seaborn provides:
Line plots: Seaborn provides several functions for creating line plots, including
lineplot, relplot, and tsplot. These functions can be used to visualize trends in
your data over time or across different variables.
Scatter plots: Seaborn's scatterplot function can be used to create scatter plots
with one or more variables. You can customize the size and color of the points
based on additional variables to create multi-dimensional scatter plots.
Bar plots: Seaborn's barplot function can be used to create bar plots with one or
more variables. You can customize the colors and orientation of the bars to create
different types of bar plots.
Heatmaps: Seaborn's heatmap function can be used to create heatmaps to
visualize relationships between two variables. You can customize the color map
and annotations to create informative heatmaps.
Pair plots: Seaborn's pairplot function can be used to create scatter plots
between all pairs of variables in your data set. This can be a useful way to explore
the relationships between multiple variables.
Joint plot: A joint plot is used to visualize the relationship between two
continuous variables and the distribution of each variable.
Histogram: A histogram is used to visualize the distribution of a single continuous
variable.
Box plot: A box plot is used to visualize the distribution of a single continuous
variable and to identify outliers.
Violin plot: A violin plot is similar to a box plot, but it provides a more detailed
view of the distribution of the data.
Regression plots: Seaborn's regplot and lmplot functions can be used to create
regression plots to visualize the relationship between two variables.

5.2 ALGORITHMS USED

5.2.1 Logistic Regression

Logistic Regression was used in the biological sciences in early twentieth century.
It was then used in many social science applications. Logistic Regression is used
when the dependent variable (target) is categorical.
For example,
To predict whether an email is spam (1) or (0)
Whether the tumor is malignant (1) or not (0)
Consider a scenario where, need to classify whether an email is spam or not. Use
linear regression for this problem, there is a need for setting up a threshold based
on which classification can be done. Say if the actual class is malignant,
predicted continuous value 0.4 and the threshold value is 0.5, the data point will
be classified as not malignant which can lead to serious consequence in real
time.
From this example, it can be inferred that linear regression is not suitable for
classification problem. Linear regression is unbounded, and this brings logistic
regression into picture. Their value strictly ranges from 0 to 1.
Uses Of Logistic Regression
Logistic regression has become particularly popular in online advertising, enabling
marketers to predict the likelihood of specific website users who will click on
particular advertisements as a yes or no percentage.
• Logistic regression can also be used in:
• Healthcare to identify risk factors for diseases and plan preventive
measures.
• Weather forecasting apps to predict snowfall and weather conditions.

• Voting apps to determine if voters will vote for a particular candidate.

Insurance to predict the chances that a policy holder will die before the term of
the policy expires based on certain criteria, such as gender, age and physical
examination.
Banking to predict the chances that a loan applicant will default on a loan or not,
based on annual income, past defaults and past debts.

5.2.2 DECISION TREE CLASSIFIER

A tree has many analogies in real life, and turns out that it has influenced a wide
area of machine learning, covering both classification and regression. In decision
analysis, a decision tree can be used to visually and explicitly represent decisions
and decision making. As the name goes, it uses a tree-like model of decisions.
Though a commonly used tool in data mining for deriving a strategy to reach a
particular goal.
A decision tree is drawn upside down with its root at the top. In the image on the
left, the bold text in black represents a condition/internal node, based on which
the tree splits into branches/ edges. The end of the branch that doesn’t split
anymore is the decision/leaf, in this case, whether the passenger died or survived,
represented as red and green text respectively.
Although, a real dataset will have a lot more features and this will just be a branch
in a much bigger tree, but you can’t ignore the simplicity of this algorithm. The
feature importance is clear and relations can be viewed easily. This methodology
is more commonly known as learning decision tree from data and above tree is
called Classification tree as the target is to classify passenger as survived or died.
Regression trees are represented in the same manner, just they predict
continuous values like price of a house. In general, Decision Tree algorithms are
referred to as CART or Classification and Regression Trees.
So, what is actually going on in the background? Growing a tree involves deciding
on which features to choose and what conditions to use for splitting, along with
knowing when to stop. As a tree generally grows arbitrarily, you will need to trim it
down for it to look beautiful. Let’s start with a common technique used for
splitting

5.2.3 RANDOM FOREST CLASSIFIER

A random forest is a machine learning technique that’s used to solve regression

and classification problems. It utilizes ensemble learning, which is a technique
that combines many classifiers to provide solutions to complex problems.
A random forest algorithm consists of many decision trees. The ‘forest’ generated
by the random forest algorithm is trained through bagging or bootstrap
aggregating. Bagging is an ensemble meta-algorithm that improves the accuracy
of machine learning algorithms.
The (random forest) algorithm establishes the outcome based on the predictions
of the decision trees. It predicts by taking the average or mean of the output from
various trees. Increasing the number of trees increases the precision of the
outcome.
A random forest eradicates the limitations of a decision tree algorithm. It reduces
the over fitting of datasets and increases precision. It generates predictions
without requiring many configurations in packages (like Scikit-learn).
Features of a Random Forest Algorithm:
● It’s more accurate than the decision tree algorithm.
● It provides an effective way of handling missing data.
● It can produce a reasonable prediction without hyper-parameter tuning.
● It solves the issue of over fitting in decision trees.
● In every random forest tree, a subset of features is selected randomly at
the node’s splitting point.

Decision trees are the building blocks of a random forest algorithm. A decision
tree is a decision support technique that forms a tree-like structure. An overview
of decision trees will help us understand how random forest algorithms work.
A decision tree consists of three components: decision nodes, leaf nodes, and a
root node. A decision tree algorithm divides a training dataset into branches,
which further segregate into other branches. This sequence continues until a leaf
node is attained. The leaf node cannot be segregated further.
The nodes in the decision tree represent attributes that are used for predicting the
outcome. Decision nodes provide a link to the leaves. The following diagram
shows the three types of nodes in a decision tree.

Fig. 5.4 Types of nodes in decision tree

The information theory can provide more information on how decision trees work.
Entropy and information gain are the building blocks of decision trees. An
overview of these fundamental concepts will improve our understanding of how
decision trees are built.
Entropy is a metric for calculating uncertainty. Information gain is a measure of
how uncertainty in the target variable is reduced, given a set of independent
variables.
The information gain concept involves using independent variables (features) to
gain information about a target variable (class). The entropy of the target variable
(Y) and the conditional entropy of Y (given X) are used to estimate the information
gain. In this case, the conditional entropy is subtracted from the entropy of Y.
Information gain is used in the training of decision trees. It helps in reducing
uncertainty in these trees. A high information gain means that a high degree of
uncertainty (information entropy) has been removed. Entropy and information gain
are important in splitting branches, which is an important activity in the
construction of decision trees.
Let’s take a simple example of how a decision tree works. To predict if a customer
will purchase a mobile phone or not. The features of the phone form the basis of
his decision. This analysis can be presented in a decision tree diagram.
The root node and decision nodes of the decision represent the features of the
phone mentioned above. The leaf node represents the final output,
either buying or notbuying. The main features that determine the choice include
the price, internal storage, and Random Access Memory (RAM). The decision tree
will appear as follows.

Fig. 5.5 Example of decision tree

Applying decision trees in random forest

The main difference between the decision tree algorithm and the random forest
algorithm is that establishing root nodes and segregating nodes is done randomly
in the latter. The random forest employs the bagging method to generate the
required prediction.
Bagging involves using different samples of data (training data) rather than just
one sample. A training dataset comprises observations and features that are used
for making predictions. The decision trees produce different outputs, depending
on the training data fed to the random forest algorithm. These outputs will be
ranked, and the highest will be selected as the final output.
Our first example can still be used to explain how random forests work. Instead of
having a single decision tree, the random forest will have many decision trees.
Let’s assume, only four decision trees. In this case, the training data comprising
the phone’s observations and features will be divided into four root nodes.
The root nodes could represent four features that could influence the customer’s
choice (price, internal storage, camera, and RAM). The random forest will split the
nodes by selecting features randomly. The final prediction will be selected based
on the outcome of the four trees.
The outcome chosen by most decision trees will be the final choice. If three trees
predict buying, and one tree predicts not buying, then the final prediction will be
buying. In this case, it’s predicted that the customer will buy the phone.

5.3 TESTING

The purpose of testing is to discover errors. Testing is the process of trying to

discover every conceivable fault or weakness in a work product. It provides a way
to check the functionality of components, sub-assemblies, assemblies and/or a
finished product It is the process of exercising software with the intent of ensuring
that the Software system meets its requirements and user expectations and does
not fail in an unacceptable manner. There are various types of test. Each test type
addresses a specific testing requirement.

5.3.1 Unit Testing

Unit testing is a crucial part of the software development process, including

machine learning models. Unit testing the detection of fake accounts on
Instagram using machine learning:

Define the Input and Output: The first step is to define the input and output of
the function or method that is testing. For example, the input could be a set of
features extracted from an Instagram account, and the output could be a binary
classification (real or fake).

Create Test Cases: Once you have defined the input and output, you can create
test cases to verify that the function or method works correctly. Test cases should
cover both positive and negative scenarios, including cases where the input is
valid and invalid.

Set Up Test Environment: You will need to set up a test environment to run the
test cases. This includes creating a testing database or file with sample data,
setting up any necessary dependencies, and configuring any environment
variables needed to run the tests.

Run Test Cases: Then run the test cases to check if the function or method is
working as expected. The tests should validate that the model accurately detects
real and fake accounts based on the provided features.

Evaluate Results: After running the tests, evaluate the results to determine if the
model is correctly identifying real and fake accounts. If any test cases fail, will
need to debug and fix the code to address the issue.

Repeat the Process: Unit testing is an iterative process, and need to repeat the
above steps as you make changes to the model or function.
.
 CHAPTER 6

RESULTS AND DISCUSSION

User can select any model among three models. Then the system will generate
accuracy for a particular selected model. Users can also view that generated
accuracy, this process will be done on the model page. The selection of the model
is done with the highest accuracy given algorithm. Logistic regression gave 89.9%
accuracy, Decision tree got 89.5% accuracy and Random forest gave the highest
accuracy of 91.8%.

Home Page
In the application ,the home page shows a welcome page of showing information
of detecting fake accounts.

Fig.6.1 Home

About Page

In the application,the about page provides the information of fake account

detection using machine learning, importance of detecting fake accounts and
necessity of detecting fake accounts on social media.
 Fig. 6.2 About

View Page

In the application, view page shows the information about the dataset with 12
attributes which is downloaded from kaggle website

Fig. 6.3 View

Model Selection Page

In the application, there is an option to choose an algorithm within that need to

selct the algorithm which gives highest accuracy which is random forest with
91.8%.
 Fig : 6.4 Model Selection

Prediction

In the application, after choosing highest accuracy given algorithm, can able to
predict an account is fake or not with the user input

Fig. 6.5 Prediction

Feature Analysis

Feature analysis is a critical step in machine learning, which involves identifying

and selecting the most relevant features in a dataset. Feature analysis helps to
improve model performance, reduce complexity, and optimize computational
efficiency.

Accuracy

Accuracy = (Number of Correct Predictions) / (Total Number of Predictions)

To illustrate this formula, consider a binary classification problem where there are
test set of 100 instances, of which 60 are labeled as class A and 40 are labeled as
class B. We train a binary classification model on the training set, and when we
evaluate the model on the test set.

Table of Accuracy

Sl No. Model Accuracy

1 Logistic Regression 89.9%

2 Random Forest 91.8%

3 Decision Tree 89.4%

Table no. 6.1 Accuracy

Accuracy for logistic regression is 88.9%.

Fig. 6.6 Logistic regression accuracy

Accuracy for random forest is 91.8%

Fig. 6.7 Random Forest accuracy

Accuracy for decision tree is 89.4%.

Fig. 6.8 Decision tree accuracy

After comparing all the three machine learning model logistic regression, random
forest, and decision tree, Random forest got the highest accuracy with 91.8%.
 CHAPTER 7

CONCLUSION
7.1 CONCLUSION

A machine-learning approach for detecting fake accounts on social media

as previously stated, three popular algorithms were compared. After the
comparison, random forest algorithm got the best accuracy. Hence, Random
Forest classifier algorithm can be considered for the rest of the project. Simply,
our project has two phases- frontend and backend. In the backend, did the data
pre- processing, model selection, prediction and taking best accuracy given
algorithm. Later, in the front end included this code in a framework and with the
user input can detect the fake account or not. Hence, concluded that detecting
fraudulent profiles depends on the features used in training the data set and
classifiers used.

7.2 FUTURE WORK

Future work in developing of detecting fake accounts on Instagram. Here are

some examples:
Larger And More Diverse Dataset: In order to increase the accuracy and
generalizability of the fake account detection model, it may be useful to test it on
a larger and more diverse dataset of Instagram users. This could involve
collecting data from users in different regions, with different types of accounts
and activity patterns.

Incorporating Additional Features: As new data becomes available, there may

be additional features that could be incorporated into the fake account detection
model to improve its accuracy. For example, Instagram's API could provide
additional data on user engagement and activity patterns, which could be used to
distinguish between real and fake accounts.

Implementing real-time detection: A potential area for future work is to implement

real-time detection of fake accounts. This could involve creating a system that
monitors user behavior in real-time and provides alerts if suspicious activity is
detected. This would be particularly useful for identifying and removing bot
accounts that are actively spamming or engaging in other malicious activities.

Investigating the use of Deep Learning: Deep learning techniques, such as

convolutional neural networks (CNNs) and recurrent neural networks (RNNs), have
been shown to be effective in a variety of machine learning applications.
Investigating the use of these techniques for fake account detection on Instagram
could be a fruitful area for future research.

Collaborating with Instagram and other social media platforms: Collaborating

with Instagram and other social media platforms could be a valuable way to
improve the effectiveness of fake account detection methods. This could involve
sharing data and insights, or working together to develop new tools and
techniques for identifying and removing fake accounts from the platform.

7.3 RESEARCH ISSUES

Data collection: Collecting and annotating a large dataset of Instagram users is a

challenging task, particularly when it comes to identifying fake accounts.
Research in this area could focus on developing more efficient and accurate
methods for data collection and annotation, as well as exploring the best
practices for balancing the need for data quality with user privacy concerns.

Feature selection and extraction: Identifying the most relevant features for fake
account detection is a crucial step in developing an effective machine learning
model. Research in this area could focus on identifying new features that can help
distinguish between real and fake accounts, as well as exploring the best
methods for feature extraction and selection.

Model selection and evaluation: Choosing the most appropriate machine

learning algorithm for fake account detection is an important decision, and there
are many different options to choose from. Research in this area could explore the
effectiveness of different machine learning algorithms for fake account detection,
as well as the best practices for model evaluation and selection.
Addressing imbalanced data: Imbalanced data is a common problem in fake
account detection, as there are typically far more real accounts than fake
accounts in any given dataset. Research in this area could focus on developing
effective methods for addressing imbalanced data, such as oversampling, under
sampling, or using cost-sensitive learning algorithms.

Ethics and privacy: Developing a fake account detection model raises important
ethical and privacy concerns, particularly when it comes to user data collection
and sharing. Research in this area could explore the best practices for balancing
the need for data privacy with the need for data quality, as well as the ethical
implications of using machine learning to detect and remove fake accounts on
social media platforms.

7.3 IMPLEMENTATION ISSUES

Data collection: Collecting a large and diverse dataset of Instagram users is a

crucial step in developing an effective fake account detection model. However,
collecting data can be time-consuming and resource-intensive, particularly when
it comes to identifying and annotating fake accounts. Implementing an efficient
and effective data collection process will be important for the success of the
project.

Feature engineering: Identifying the most relevant features for fake account
detection is another important step in developing an effective machine learning
model. However, feature engineering can be a complex process that requires
careful consideration of both the features themselves and the algorithms used to
extract them. Implementing an effective feature engineering process will be key to
developing an accurate and efficient model.

Model selection and tuning: Choosing the most appropriate machine learning
algorithm and tuning its parameters is another important step in developing an
effective fake account detection model. Different algorithms and parameter
settings can have a significant impact on the performance of the model, and the
process of selecting and tuning the model can be time-consuming and require a
deep understanding of machine learning techniques.

Deployment and scalability: Once a fake account detection model has been
developed, deploying it to a production environment can be a complex process.
The model may need to be integrated with existing systems and software, and it
may need to be tested and optimized for scalability and performance.
Implementing an effective deployment and scalability plan will be important for
ensuring that the model can be used effectively in a real-world setting.

User privacy and ethical considerations: Developing a fake account detection

model raises important ethical and privacy concerns, particularly when it comes
to user data collection and sharing. Implementing appropriate privacy policies
and procedures, and ensuring that the project adheres to ethical guidelines and
regulations, will be important for ensuring that the project is both effective and
responsible.
REFERENCES

[1] B. Erşahin, Ö. Aktaş, D. Kılınç, and C. Akyol, "Twitter fake account

detection," 2017 International Conference on Computer Science and
Engineering (UBMK), Antalya, Turkey, 2017, pp. 388-392, doi:
10.1109/UBMK.2017.8093420.
[2] A. Gupta and R. Kaushal, "Towards detecting fake user accounts in
Facebook," 2017 ISEA Asia Security and Privacy (ISEASP), Surat, India,
2017, pp. 1-6, doi: 10.1109/ISEASP.2017.7976996.
[3] Y. -C. Chen and S. F. Wu, "Fake Buster: A Robust Fake Account Detection
by Activity Analysis," 2018 9th International Symposium on Parallel
Architectures, Algorithms and Programming (PAAP), Taipei, Taiwan, 2018,
pp. 108-110, doi: 10.1109/PAAP.2018.00026.
[4] E. Van Der Walt and J. Eloff, "Using Machine Learning to Detect Fake
Identities: Bots vs Humans," in IEEE Access, vol. 6, pp. 6540-6549, 2018,
doi: 10.1109/ACCESS.2018.2796018.
[5] L. Zhang, Z. Wu and J. Cao, "Detecting Spammer Groups from Product
Reviews: A Partially Supervised Learning Model," in IEEE Access, vol. 6,
pp. 2559-2568, 2018, doi: 10.1109/ACCESS.2017.2784370.
[6] S. Khaled, N. El-Tazi and H. M. O. Mokhtar, "Detecting Fake Accounts on
Social Media," 2018 IEEE International Conference on Big Data (Big Data),
Seattle, WA, USA, 2018, pp. 3672-3681, doi:
10.1109/BigData.2018.8621913.
[7] Z. Alom, B. Carminati and E. Ferrari, "Detecting Spam Accounts on
Twitter," 2018 IEEE/ACM International Conference on Advances in Social
Networks Analysis and Mining (ASONAM), Barcelona, Spain, 2018, pp.
1191-1198, doi: 10.1109/ASONAM.2018.8508495.
[8] N. Singh, T. Sharma, A. Thakral and T. Choudhury, "Detection of Fake
Profile in Online Social Networks Using Machine Learning," 2018
International Conference on Advances in Computing and Communication
Engineering (ICACCE), Paris, France, 2018, pp. 231-234, doi:
10.1109/ICACCE.2018.8441713.
[9] S. Singhal, R. R. Shah, T. Chakraborty, P. Kumaraguru and S. Satoh, "Spot
Fake: A Multi-modal Framework for Fake News Detection," 2019 IEEE Fifth
 International Conference on Multimedia Big Data (BigMM), Singapore,
2019, pp. 39-47, doi: 10.1109/BigMM.2019.00-44.
[10] P. Sowmya and M. Chatterjee, "Detection of Fake and Clone
accounts in Twitter using Classification and Distance Measure
Algorithms," 2020
International Conference on Communication and Signal Processing
(ICCSP), Chennai, India, 2020, pp. 0067-0070, doi:
10.1109/ICCSP48568.2020.9182353.
[11] Z. Shahbazi and Y. -C. Byun, "Fake Media Detection Based on
Natural Language Processing and Blockchain Approaches," in IEEE
Access, vol. 9, pp. 128442-128453, 2021, doi:
10.1109/ACCESS.2021.3112607.
[12] L. P, S. V, V. Sasikala, J. Arunarasi, A. R. Rajini and N. Nithiya, "Fake
Profile Identification in Social Network using Machine Learning and
NLP," 2022 International Conference on Communication, Computing and
Internet of Things (IC3IoT), Chennai, India, 2022, pp. 1-4, doi:
10.1109/IC3IOT53935.2022.9767958.
M. J. Ekosputra, A. Susanto, F. Haryanto and D. Suhartono, "Supervised
Machine Learning Algorithms to Detect Instagram Fake Accounts," 2021
4th International Seminar on Research of Information Technology and
Intelligent Systems (ISRITI), Yogyakarta, Indonesia, 2021, pp. 396-400, doi:
10.1109/ISRITI54043.2021.9702833.
[13] K. Anklesaria, Z. Desai, V. Kulkarni and H. Balasubramaniam, "A Survey
on Machine Learning Algorithms for Detecting Fake Instagram Accounts,"
2021 3rd International Conference on Advances in Computing,
Communication Control and Networking (ICAC3N), Greater Noida, India,
2021, pp. 141-144, doi: 10.1109/ICAC3N53548.2021.9725724.
[14] M. M. Hassan Sohan, M. M. Khan, I. Nanda and R. Dey, "Fake Product
Review Detection Using Machine Learning," 2022 IEEE World AI IoT
Congress (AIIoT), Seattle, WA, USA, 2022, pp. 527-532, doi:
10.1109/AIIoT54504.2022.9817271.
[15] Praveena, M.A., Christy, A., Helen, L.S., Krishna, R.S. and Nandini,
D.U., 2021, March. Examining and Predicting Helpfulness of reviews based
on Naive Bayes. In Journal of Physics: Conference Series (Vol. 1770, No. 1,
p. 012021). IOP Publishing.
 APPENDIX

A. SOURCE CODE :

Home.html

<!DOCTYPEhtml>
<html>
<head>
<!-- Basic -->
<meta charset="utf-8" />
<meta http-equiv="X-UA-Compatible" content="IE=edge" />
<!-- Mobile Metas -->
<meta name="viewport" content="width=device-width, initial-scale=1,
shrink-to-fit=no" />
<!-- Site Metas -->
<meta name="keywords" content="" />
<meta name="description" content="" />
<meta name="author" content="" />

<title>Fake Account</title>
<link rel="icon" href="static/images/1.png" type="image/icon type">

<!-- slider stylesheet -->

<link rel="stylesheet" type="text/css"
href="https://cdnjs.cloudflare.com/ajax/libs/OwlCarousel2/2.1.3/assets/ow
l.carousel.min.css" />

<!-- bootstrap core css -->

<link rel="stylesheet" type="text/css" href="static/css/bootstrap.css" />

<!-- fonts style -->

<link
href="https://fonts.googleapis.com/css?family=Poppins:400,700&display=s
wa p" rel="stylesheet" />
<!-- Custom styles for this template -->
<link href="static/css/style.css" rel="stylesheet" />
<!-- responsive style -->
<link href="static/css/responsive.css" rel="stylesheet" />
</head>

<body>
<div class="hero_area">
<!-- header section strats -->
<header class="header_section">
<div class="container">
<nav class="navbar navbar-expand-lg custom_nav-container pt-3">
<a class="navbar-brand" href="{{url_for('index')}}">
<img src="static/images/1.png" alt="" style="width: 200px;" /><span>
</span>
</a>
<button class="navbar-toggler" type="button" data-toggle="collapse" data-
target="#navbarSupportedContent"
aria-controls="navbarSupportedContent" aria-expanded="false" aria-
label="Toggle navigation">
<span class="navbar-toggler-icon"></span>
</button>
Index.html

<!DOCTYPE html>
<html>
<head>
<!-- Basic -->
<meta charset="utf-8" />
<meta http-equiv="X-UA-Compatible" content="IE=edge" />
<!-- Mobile Metas -->
<meta name="viewport" content="width=device-width, initial-scale=1,
shrink-to-fit=no" />
<!-- Site Metas -->
<meta name="keywords" content="" />
<meta name="description" content="" />
<meta name="author" content="" />

<title>Fake Account</title>
<link rel="icon" href="static/images/1.png" type="image/icon type">
<!-- slider stylesheet -->
<link rel="stylesheet" type="text/css"
href="https://cdnjs.cloudflare.com/ajax/libs/OwlCarousel2/2.1.3/assets/ow
l.carousel.min.css" />

<!-- bootstrap core css -->

<link rel="stylesheet" type="text/css" href="static/css/bootstrap.css" />

<!-- fonts style -->

<link
href="https://fonts.googleapis.com/css?family=Poppins:400,700&display=s
wa p" rel="stylesheet" />
<!-- Custom styles for this template -->
<link href="static/css/style.css" rel="stylesheet" />
<!-- responsive style -->
<link href="static/css/responsive.css" rel="stylesheet" />
</head>

<body>
<div class="hero_area">
<!-- header section strats -->
<header class="header_section">
<div class="container">
<nav class="navbar navbar-expand-lg custom_nav-container pt-3">
<a class="navbar-brand" href="{{url_for('index')}}">
<img src="static/images/1.png" alt="" style="width: 200px;" /><span>
</span>
</a>
<button class="navbar-toggler" type="button" data-toggle="collapse" data-
target="#navbarSupportedContent"
aria-controls="navbarSupportedContent" aria-expanded="false" aria-
label="Toggle navigation">
<span class="navbar-toggler-icon"></span>
</button>

<div class="collapse navbar-collapse" id="navbarSupportedContent">

<div class="d-flex ml-auto flex-column flex-lg-row align-items-center">
<ul class="navbar-nav ">
<li class="nav-item active">
<a class="nav-link" href="{{url_for('index')}}">Home <span class="sr-
only">(current)</span></a>
</li>
View.html
<!DOCTYPE html>
<html>

<head>
<!-- Basic -->
<meta charset="utf-8" />
<meta http-equiv="X-UA-Compatible" content="IE=edge" />
<!-- Mobile Metas -->
<meta name="viewport" content="width=device-width, initial-scale=1,
shrink-to-fit=no" />
<!-- Site Metas -->
<meta name="keywords" content="" />
<meta name="description" content="" />
<meta name="author" content="" />

<title>Fake Account</title>
<link rel="icon" href="static/images/1.png" type="image/icon type">

<!-- slider stylesheet -->

<link rel="stylesheet" type="text/css"

href="https://cdnjs.cloudflare.com/ajax/libs/OwlCarousel2/2.1.3/assets/ow
l.carousel.min.css" />

<!-- bootstrap core css -->

<link rel="stylesheet" type="text/css" href="static/css/bootstrap.css" />

<!-- fonts style -->

<link
href="https://fonts.googleapis.com/css?family=Poppins:400,700&display=s
wa p" rel="stylesheet" />
<!-- Custom styles for this template -->
<link href="static/css/style.css" rel="stylesheet" />
<!-- responsive style -->
<link href="static/css/responsive.css" rel="stylesheet" />
</head>

<body>
<div class="hero_area">
<!-- header section strats -->
<header class="header_section">
<div class="container">
<nav class="navbar navbar-expand-lg custom_nav-container pt-3">
<a class="navbar-brand" href="{{url_for('index')}}">
<img src="static/images/1.png" alt="" style="width: 200px;" /><span>

</span>
</a>
<button class="navbar-toggler" type="button" data-toggle="collapse" data-
target="#navbarSupportedContent"
aria-controls="navbarSupportedContent" aria-expanded="false"
aria-label="Toggle navigation">
<span class="navbar-toggler-icon"></span>
</button>
Model.html
<!DOCTYPE html>
<html>
<head>
<!-- Basic -->
<meta charset="utf-8" />
<meta http-equiv="X-UA-Compatible" content="IE=edge" />
<!-- Mobile Metas -->
<meta name="viewport" content="width=device-width, initial-scale=1,
shrink-to-fit=no" />
<!-- Site Metas -->
<meta name="keywords" content="" />
<meta name="description" content="" />
<meta name="author" content="" />
<title>Fake Account</title>
<link rel="icon" href="static/images/1.png" type="image/icon type">
<!-- slider stylesheet -->
<link rel="stylesheet"
type="text/css"href="https://cdnjs.cloudflare.com/ajax/libs/OwlCarousel2/
2.1.3/assets/owl.carousel.min.css" />
<!-- bootstrap core css -->
<link rel="stylesheet" type="text/css" href="static/css/bootstrap.css" />
<!-- fonts style -->
<link
href="https://fonts.googleapis.com/css?family=Poppins:400,700&display=s
wa p" rel="stylesheet" />
<!-- Custom styles for this template -->
<link href="static/css/style.css" rel="stylesheet" />
<!-- responsive style -->
<link href="static/css/responsive.css" rel="stylesheet" />
</head>
<body>
<div class="hero_area">
<!-- header section strats -->
<header class="header_section">
<div class="container">
<nav class="navbar navbar-expand-lg custom_nav-container pt-3">
<a class="navbar-brand" href="{{url_for('index')}}">
<img src="static/images/1.png" alt="" style="width: 200px;" /><span>
</span>
</a>
<button class="navbar-toggler" type="button" data-toggle="collapse" data-
target="#navbarSupportedContent"
aria-controls="navbarSupportedContent" aria-expanded="false" aria-
label="Toggle navigation">
<span class="navbar-toggler-icon"></span>
</button>
<div class="collapse navbar-collapse" id="navbarSupportedContent">
<div class="d-f avbar-nav ">
<li class="nav-item active">
<a class="nav-link" href="{{url_for('index')}}">Home <span class="sr-
only">(current)</span></a>
</li>
<li class="nav-item">
<a class="nav-link" href="{{url_for('about')}}">About</a>
</li>
<li class="nav-item">
<a class="nav-link" href="{{url_for('view')}}">View </a>
</li
Prediction.html

<!DOCTYPE html>
<html>

<head>
<!-- Basic -->
<meta charset="utf-8" />
<meta http-equiv="X-UA-Compatible" content="IE=edge" />
<!-- Mobile Metas -->
<meta name="viewport" content="width=device-width, initial-scale=1,
shrink-to-fit=no" />
<!-- Site Metas -->
<meta name="keywords" content="" />
<meta name="description" content="" />
<meta name="author" content="" />

<title>Fake Account</title>
<link rel="icon" href="static/images/1.png" type="image/icon type">
<!-- slider stylesheet -->
<link rel="stylesheet"
type="text/css"href="https://cdnjs.cloudflare.com/ajax/libs/OwlCarousel2/
2.1.3/assets/owl.carousel.min.css" />
<!-- bootstrap core css -->
<link rel="stylesheet" type="text/css" href="static/css/bootstrap.css" />
<!-- fonts style -->
<link
href="https://fonts.googleapis.com/css?family=Poppins:400,700&display=s
wa p" rel="stylesheet" />
<!-- Custom styles for this template -->
<link href="static/css/style.css" rel="stylesheet" />
<!-- responsive style -->
<link href="static/css/responsive.css" rel="stylesheet" />
</head>

<body>
<div class="hero_area">
<!-- header section strats -->
<header class="header_section">
<div class="container">
<nav class="navbar navbar-expand-lg custom_nav-container pt-3">
<a class="navbar-brand" href="{{url_for('index')}}">
<img src="static/images/1.png" alt="" style="width: 200px;" /><span>
</span>
</a>
<button class="navbar-toggler" type="button" data-toggle="collapse" data-
target="#navbarSupportedContent"
aria-controls="navbarSupportedContent" aria-expanded="false" aria-
label="Toggle navigation">
<span class="navbar-toggler-icon"></span>
</button>
<div class="collapse navbar-collapse" id="navbarSupportedContent">
<div class="d-flex ml-auto flex-column flex-lg-row align-items-center">
<ul class="navbar-nav ">
<li class="nav-item active">
<a class="nav-link" href="{{url_for('index')}}">Home <span class="sr-
App.py
import pandas as pd
from flask import *
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from sklearn.tree import DecisionTreeClassifier from
sklearn.ensemble import RandomForestClassifier from
sklearn.linear_model import LogisticRegression app =
Flask( name )
@app.route('/')
def index():
return render_template('index.html')
@app.route('/about')
def about():
return render_template('about.html')

# @app.route('/load',methods=["GET","POST"])
# def load():
# global df, dataset
# if request.method == "POST":
# data = request.files['data']
# df = pd.read_csv(data)
# dataset = df.head(100)
# msg = 'Data Loaded Successfully'
# return render_template('load.html', msg=msg)
# return render_template('load.html')
@app.route('/view')
def view():
global df, dataset
df = pd.read_csv('data.csv')
dataset = df.head(100)
return render_template('view.html', columns=dataset.columns.values,
rows=dataset.values.tolist())
@app.route('/model',methods=['POST','GET'])
def model():
if request.method=="POST":
data = pd.read_csv('data.csv')
data.head()
x=data.iloc[:,:-1]
y=data.iloc[:,-1]

x_train,x_test,y_train,y_test =
train_test_split(x,y,test_size=0.3,stratify=y,random_state=42)print('cccc
ccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc')
s=int(request.form['algo'])
if s==0:
return render_template('model.html',msg='Please Choose an Algorithm to
Train')
elif
s==1:print('aaaaaaaaaaaaaaaaaaaaabbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb')
from sklearn.linear_model import LogisticRegression
lr = LogisticRegression()
lr=lr.fit(x_train,y_train)
y_pred = lr.predict(x_test)
acc_lr = accuracy_score(y_test,y_pred)*100
print('aaaaaaaaaaaaaaaaaaaaaaaaa')
msg = 'The accuracy obtained by Logistic Regression is ' +
str(acc_lr) + str('%')
B. SCREEN SHOTS

Ipnyb.py
 C. RESEARCH PAPER :

FAKE ACCOUNT DETECTION ON SOCIAL

MEDIA USING RANDOM FORESTCLASSIFIER

Kancharla Venkata Nikhitha1 Karnati Bhavya2 Dr. D. Usha Nandini3

Student Student Associate Professor
Department of Computer Department of Computer Department of Computer
Science Scienceand Engineering, Scienceand Engineering,
and Engineering, Sathyabama Institute of Science Sathyabama Institute of Science
Sathyabama Institute of Science and Technology, and Technology,
and Chennai, India. Chennai, India.
Technology, [email protected] [email protected]
Chennai, India. m
[email protected]

Fraudulent accounts are one of the

ABSTRACT–As the growth of online social networks
downsides of Online Social Networks. We
increases everyone is associated and linked with social can check an account is fake or not by
media. A massive amount of personal data is being following parameters, the number of
attacked and stolen by cyber attackers. These followers, and posts, or the characteristics
fraudulent profiles also spread negative publicity, of the posted information such as the
false news, and various malicious programs one of the number of likes, views, comments, and
best ways is to detect these fake accounts, not only the shares. By knowing this, many fake
accounts are formed and started
best way but also it is the first step to stopping the
purchasing comments and increasing the
negative news and false rumours. There are many number of followers to become popular.
efficient ways and techniques to detect these fake These are particularly created to commit
accounts. However, these techniques are based on or fraudulent acts such as disseminating
depend on the account features. In this paper, main misleading information and spreading
aim is to examine three machine learning algorithms malware. While some are created to gain
that are random forest, logistic regression, and followers to become popular, spam
comments and likes. In social media, one
decision tree. Now, progressing to compare the level of
individual creates many profiles with
efficiency in the three mentioned methodologies and distinct identities, email addresses, and
got the highest accuracy for random forest classifier. phone numbers. The majority of them are
likely fake accounts. There are two sorts
Keywords: Social media, Regression, Fake accounts, of fake accounts: duplicate accounts and
Identify theft fraudulent accounts. Users create multiple
accounts as secondary accounts to
I. INTRODUCTION promote their e-businesses. People and
influencers establish duplicate accounts to
In recent times, some of the most well-known social enhance their businesses and spread
network sites like Instagram, Facebook and Twitter useful information. This type of account is
became integral parts of daily life. Social network sites not harmful and does not violate social
are used by people for e- commerce, entertainment, media terms and conditions. Users create
information and idea sharing keeping in contact with fake accounts mainly to spread negativity,
long-lost friends and finding new acquaintances. Users of false news, hate speech, and impersonate
OSN can post pictures and videos, leave comments and others; these accounts might be
also provide likes to pictures that have been shared. considered dangerous and called as false
accounts.
This work primarily concentrated on
detecting if an Instagram account is phony
or real. This detection is based on some
parameters such as username, private or
public account, and number of followers,
posts, follows and profile pictures. We can
detect a user's activity using this
attributes.The user’s behaviour on
Instagram is determined by a collection of
12 features. Three
machine learning techniques are used to distinguish intelligence to identify Phony identities.
between genuine and fraudulent accounts. The author of this research clarified how
to spot false profiles made by both bots
and humans. In this study, the fraudulent
II. LITERATURE REVIEW
accounts made on social networks by
Detection of fake accounts is one of the major issues that people and bots were compared. Finally, it
should be solved as soon as possible. Though various can be said that while the features
methods have been already making an impact, these all employed to detect the bots were
methods are not completely accurate. Inspired by various successful, they weren’t entirely effective.
research papers on detection, in this whole literature
review section, we are going to discuss various applied Lu Zhang et.al [5], suggested the PSGD, a
methodologies and approaches. partially supervised model, to identify
scammer groups. As a scammer group
Ersahin et.al [1], provided a categorization technique to detector, a classifier is studied using
find Twitter's Phony accounts. And pre-processed our PU-Learning via PSGD. The suggested
dataset by applying the Entropy Minimization PSGD is effective and surpasses
Discretization (EMD) method under supervision to cutting-edge spammer detection
numerical characteristics, and then they examined the techniques, according to tests on a
output of the Naive Bayes algorithm. By merely real-world data set from
pre-processing their dataset using the n discretization Amazon.cn.Bucket online media, they
strategy on chosen characteristics, they were able to believe it to be a significant and extremely
improve the accuracy with Nave Bayes from85.55% promising outcome.
to.90.41% for a method for identifying bogus accounts on
Twitter. Only by applying the numerical information from Sarah Khaled et.al [6], investigated
provide a safe platform that can forecast and spot fake effective methods for spotting bots and
news in social media networks. Phony accounts on the Twitter network.
An innovative technique called SVM-NN
Aditi Gupta et.al [2], used the Facebook Graph API, and is recommended in this to offer an
collected Facebook user feeds. The intricate privacy effective detection for such profiles. Even
controls on Facebook made it extremely difficult to though the recommended approach uses
gather data. On a minimal dataset that included their fewer features, it can still be categorized
node, their acquaintances in their social neighbourhood, with around ninety- eight percent
as well as a collection of manually recognized spam accuracy. In comparison to the other two
accounts, they performed the most widely used classifiers, the newly suggested method
supervised machine learning classification approaches. performs better in terms of accuracy
They assessed these classifiers’ performance to identify across all feature sets. The accuracy of the
the top classifiers that produced high detection rates and correlation feature set's records is
their capacity to identify Phony Facebook accounts. astounding. This is because it chooses the
Additionally, they looked into the feature set to identify best set of traits rather than combining all
the characteristics that Facebook fake profiles are most features in a linear combination.
successfully detected by.
Zulfikar Alom et.al [7], investigated the
Yeh-Cheng Chen et.al [3], introduced a novel, efficient nature of spam users on the Twitter
technique for detecting bogus profiles using machine platform to develop an existing
learning. Their method based on account-by-account spam-detecting mechanism. In this paper,
behavioural analysis evaluate whether an account is it is designed a new mechanism and a
genuinely fake or not accurately, as opposed to utilizing more robust set of features to detect
graph-based or manually predicted that just cover basic spammers on Twitter. In this Random
facts. On real-world data, their detection models work Forest classifier is used which gives a
admirably, and their findings indicate that the models are better result compared to other
not over fitting. methodologies. From this, they can plan
to build a more effective model which
Estee van der walt et.al, [4] employed artificial classifies various types of spammers
within different social networks.
Naman Singh et.al [8], proposed that on
online networking platforms, more bogus
accounts are being created for nefarious
purposes. They create a forged human
profile in order to successfully detect,
identify, and remove fake profiles. For
bots, ML models employed a variety of
factors to determine how many people an
account has on each platform’s friend list
as followers. It is not possible to contrast
between Phony profiles made by humans
and cyborgs. By using a data set with
Phony profiles and classifying them as
fake or real, they can contrast fake and
real accounts.

Shivangi Singhal et.al [9], formally introduced

unveil Spot Fake a multi-modal platform for identifying accomplished by using a methodical
bogus news. Without considering any other subtasks, approach that includes cleaning,
their suggested approach detects bogus news. It makes pre-processing, feature selection, and
use of article textual and graphic components. Text model training. After that, all the
characteristics were learned using language models (such employed algorithms—Random Forest,
as BERT), and image features were learned using AdaBoost, MLP, SGD, and Artificial
VGG-19 pre-trained on Image Net dataset. Twitter and neural network—are compared on the
Weibo, two publicly accessible databases, are used in all basis of different evaluation parameters.
of the studies. According to the study, Random Forest
classifier performs better.
Sowmya P et.al [10], one serious issue involves
generating duplicate profile users using the data of Md Mahadi Hassan Sohan et.al [15],
existing users. A collection of principles that may showed the importance of comments and
differentiate real or fake are used to detect Phony profiles. how they affect almost every aspect of
A detection method that can find fraudulent or clone social media data. People's perspectives
profiles on online networking platforms like Twitter has are significantly influenced by reviews.
been presented. Hence, identifying fake reviews is an
exciting study area. A method for spotting
Zeinab Shahbazi et.al [11], proposed an integrated system fake reviews using machine learning was
for multiple block chains and NLP functions to leverage disclosed in the study. It detects both
ML models to find false news and better anticipate bogus review characteristics and reviewer
user accounts and posts. For this workflow, behaviour. The food review dataset was
Reinforcement learning methodology is applied. The used to evaluate the proposed method.
operation of the decentralized block chain architecture, The technique is semi-supervised, with
which offers the framework of digital content several classifiers being used. In the fake
authorization proof, improved the security of this review detection procedure, the findings
platform. The idea behind this approach is to provide a reveal that the Random Forest classifier
safe platform that can forecast and spot fake news on beats another classifier. Moreover, the
social media networks. findings imply that including in viewers'
behavioural traits raises the F-score by
Latha P et.al [12], stated different categorization 55.5% and the overall accuracy to 97.7%.
techniques have been used in the past to identify Phony
online networking media accounts. However, they must III. PROPOSED SYSTEM:
improve their ability to spot Phony accounts on these
websites. To improve the accuracy rate of identifying In this article, an effective method to
bogus accounts, they use ML technologies and NLP in detect fake accounts on online social
our study. The Random Forest tree classifying algorithm networks is proposed. This work may be
was chosen. regarded as a valuable system since it aids
in lowering the restrictions brought on by
Michael Jonathan Ekosputra et.al [13], study's findings conventional and other existing
make it evident that ml may be used to detect fake methodologies. The goal of this project is
profiles; the models that have been examined in this to provide an efficient and dependable
article include Logistic Regression, Bernoulli Naive system for precisely detecting fake. A
Bayes, Random Forest, svm, and ANN. Any changes or potent algorithm in a Python-based
additions to features will have an impact on each model's framework to develop this system is
accuracy. According to the research, adding parameters to employed. The process is depicted in the
a model will almost certainly make it more accurate than picture below.
a model with no parameters or a default form. Based on The steps for executing the proposed work:
the results of the research's initial trial, the Random
8. The required packages should be installed.
Forest algorithm achieves the best outcome with a
startling accuracy of 0.92. 9. Establishing the issue association.
Karishma Anklesaria et.al [14], used different classifiers 10. Create a framework-based UI.
to classify the profile as fake or real using database of
user accounts. This is 11. Upload CSV file

12. View data

13. Select model

14. Output to predict.

 will be turned into numerical values and
removing unwanted columns. The null
values are replaced by using imputation
technique. Then splits the dataset into
training and testing set. The target variable
or Dependent variable undergoes training
and independent variables will go to
testing. Next is model building. Model
buildings mean choosing a suitable
algorithm and train it as per the
requirements. Different models like
decision tree, Random Forest, and Logistic
Regression are considered, and train the
models by importing those modules from
Sklearn. Then, train the models using the
training dataset with fit( ) function. After
the model building, apply that model to
Figure 1: Flow Chart make predictions and predict the response
for test dataset. Then evaluate the model’s
This proposed work is completely based on Machine performance by importing different metrics
Learning approach which detects whether the profile from sklearn metrics. Random forest got
isfake or real one. To do this, three well-known the highest accuracy with 90%. So, this is
methodologies were compared. This proposed work has how the system will do preprocessing and
two phases, the first phase is frontend – which creates a User can view the dataset.
platform another is backend – a model is created and the ALGORITHMS USED:
main part of the project exists. Logistic regression:
The main part of the proposed work is the backend. Logistic Regression is commonly used
Here, using flask framework an application is created machine learning algorithm for binary
that detects whether an account is fake or not. In this classification problem, that have two class
aspect, the data is collected and preprocessed. Then the values and include predictions like “yes”
data splitting takes place. Now the data gets trained and or “no” and “a” or ”b”. It is used when the
need to build the model. After the model building with dependent variable (target) is categorical.
the help of the model data gets tested. And finally the For example, to predict whether an email is spam
prediction is done. (1) or (0) whether the account is fake (1) or
Firstly, download the dataset from Kaggle which is raw not (0).
data. The size of the dataset is 697 with 12 attributes Decision tree :
such as profile picture, length of the username, full name Decision tree algorithms are referred as
words, length of the full name, the description provided CART or Classification And Regression
by the user known as bio, any external URLs provided, Techniques. It is utilized in both regression
private or public accounts, number of posts posted, and classification. A decision tree is drawn
number of followers and number of accounts the user upside-down with its root at the top. As the
follows. name implies, it uses a tree-like model of
decision.
Random forest :
A random forest classifier is a machine
learning technique which is used for
solving regression and classification
problems. It makes use of ensemble
learning, a technique that combines
multiple classifiers to solve complicated
problem. Random forest classifier provides
an effective way of handling missing data
Figure 2: Dataset and it is more accurate than the decision
In the application user first uploads a CSV file of the tree algorithm and also solves issue of
dataset then the system takes that file and undergoes overfitting in decision trees.
preprocessing, which means Importing libraries such as
pandas, and NumPy. Import dataset, cleaning of null IV EXPERIMENTAL RESULTS:
values, categorical values
The user can select any model among three
models. Then the system will generate
accuracy for a particular selected model.
Users can also view that generated
accuracy, this process will be done on the
model page. The selection of the model is
done with the highest accuracy given
algorithm. Logistic regression gave 89.9%
accuracy, Decision tree got 89.5%
accuracy and
Random forest gave the highest accuracy of 90%. Conference on Computer Science
and Engineering (UBMK), Antalya,
Turkey, 2017, pp.
388-392, doi:
10.1109/UBMK.2017.8093420.
[17] A. Gupta and R. Kaushal, "Towards
detecting fake user accounts in
Facebook," 2017 ISEA Asia Security
and Privacy (ISEASP), Surat, India,
2017, pp.
1-6, doi:
10.1109/ISEASP.2017.7976996.
[18] Y. -C. Chen and S. F. Wu, "Fake
Buster: A Robust Fake Account
Detection by Activity Analysis,"
Figure 3: Model Accuracy
2018 9th International Symposium
The system will generate a graph that compared the
accuracy of three models and the user can view that on Parallel Architectures,
graph. So, select the model and can start the prediction Algorithms and Programming
process with the user input we predict whether an (PAAP), Taipei, Taiwan, 2018,
account is fraudulent or not. pp. 108-110,
doi:
10.1109/PAAP.2018.00026.
[19] E. Van Der Walt and J. Eloff, "Using
Machine Learning to Detect Fake
Identities: Bots vs Humans," in
IEEE Access, vol. 6, pp. 6540-6549,
2018, doi:
10.1109/ACCESS.2018.2796018.
[20] L. Zhang, Z. Wu and J. Cao,
"Detecting Spammer Groups from
Figure 4: Prediction Product Reviews: A Partially
Supervised Learning Model," in
V CONCLUSION: IEEE Access, vol. 6, pp. 2559-2568,
2018, doi:
In this paper, we proposed a machine-learning 10.1109/ACCESS.2017.2784370.
approach for detecting fake accounts on social media [21] S. Khaled, N. El-Tazi and H. M.
as previously stated, three popular algorithms were O. Mokhtar, "Detecting Fake
compared. After the comparison, random forest
Accounts on Social Media," 2018
algorithm got the best accuracy. Hence, Random
Forest classifier algorithm can be considered for the IEEE International Conference on
rest of the project.Simply, our project has two phases- Big Data (Big Data), Seattle, WA,
frontend and backend. In the backend, did the data USA, 2018, pp. 3672-3681,
pre-processing, model selection, prediction and taking
best accuracy given algorithm. Later, in the front end doi:
included this code in a framework and with the user
10.1109/BigData.2018.8621913.
input can detect the fake account or not. Hence,
concluded that detecting fraudulent profiles depends [22] Z. Alom, B. Carminati and E.
on the features used in training the data set and Ferrari, "Detecting Spam Accounts
classifiers used. on Twitter," 2018 IEEE/ACM
International Conference on
VI REFERENCE: Advances in Social Networks
Analysis and Mining (ASONAM),
[16]B. Erşahin, Ö. Aktaş, D. Kılınç, and C. Barcelona, Spain, 2018, pp.
Akyol, "Twitter fake account detection," 2017
1191-1198,
International
doi:
10.1109/ASONAM.2018.8508495.
[23] N. Singh, T. Sharma, A. Thakral and
T. Choudhury, "Detection of Fake
Profile in Online Social Networks
Using Machine Learning," 2018
International Conference on
Advances in Computing and
Communication Engineering
(ICACCE), Paris, France,
 2018, pp. 231-234, doi: doi:
10.1109/ICACCE.2018.8441713. 10.1109/ICAC3N53548.2021.9
[24] S. Singhal, R. R. Shah, T. Chakraborty, P. 725724.
Kumaraguru and S. Satoh, "Spot Fake: A [30] M. M. Hassan Sohan, M. M.
Multi-modal Framework for Fake News Khan, I. Nanda and R. Dey,
Detection," 2019 IEEE Fifth International "Fake Product Review Detection
Conference on Multimedia Big Data (BigMM), Using Machine Learning," 2022
Singapore, 2019, pp. 39-47, doi: IEEE World AI IoT Congress
10.1109/BigMM.2019.00-44. (AIIoT), Seattle, WA, USA,
[25] P. Sowmya and M. Chatterjee, "Detection of 2022, pp. 527-532, doi:
Fake and Clone accounts in Twitter using 10.1109/AIIoT54504.2022.9817
Classification and Distance Measure 271.
Algorithms," 2020 International Conference [31] Praveena, M.A., Christy, A.,
on Communication and Signal Helen, L.S., Krishna, R.S. and
Processing (ICCSP), Chennai, India, 2020, pp. Nandini, D.U., 2021, March.
0067-0070, doi: Examining and Predicting
10.1109/ICCSP48568.2020.9182353. Helpfulness of reviews based on
[26] Z. Shahbazi and Y. -C. Byun, "Fake Media Naive Bayes. In Journal of
Detection Based on Natural Language Physics: Conference Series (Vol.
Processing and Blockchain Approaches," in 1770, No. 1, p. 012021). IOP
IEEE Access, vol. 9, pp. 128442-128453, Publishing.
2021, doi: 10.1109/ACCESS.2021.3112607.
[27] L. P, S. V, V. Sasikala, J. Arunarasi, A. R. Rajini
and N. Nithiya, "Fake Profile Identification in
Social Network using Machine Learning and
NLP," 2022 International Conference on
Communication, Computing and Internet of
Things (IC3IoT), Chennai, India, 2022,
pp. 1-4, doi:
10.1109/IC3IOT53935.2022.9767958.
[28] M. J. Ekosputra, A. Susanto, F. Haryanto and D.
Suhartono, "Supervised Machine Learning
Algorithms to Detect Instagram Fake Accounts,"
2021 4th International Seminar on Research of
Information Technology and Intelligent Systems
(ISRITI), Yogyakarta, Indonesia, 2021,
pp. 396-400, doi:
10.1109/ISRITI54043.2021.9702833.
[29] K. Anklesaria, Z. Desai, V. Kulkarni and H.
Balasubramaniam, "A Survey on Machine
Learning Algorithms for Detecting Fake
Instagram Accounts," 2021 3rd International
Conference on Advances in Computing,
Communication Control and Networking
(ICAC3N), Greater Noida, India, 2021, pp.
141-144,