Full Report Credict Card Fraud

ABSTRACT
This project aims to develop a dependable system that effectively detects fraudulent
transactions and ensures the authentication of credit card users. With the surge in credit card usage,
the occurrence of fraudulent activities has become a pressing concern, necessitating the development
of an advanced system capable of identifying and preventing such illicit actions. The system will
employ various machine learning algorithms, including decision trees and random forests, to analyze
transactional data and detect suspicious activities. By training on a dataset containing information
about past fraudulent activities, the system will learn patterns and recognize similar fraudulent
transactions. Additionally, diverse authentication methods, such as one-time passwords and security
questions, will be implemented to verify the legitimacy of credit card users. The implementation of
this system is expected to enhance the security of credit card transactions by proactively detecting
and authenticating fraud, resulting in substantial savings for consumers and financial institutions.
Overall, this project's focus is on creating a reliable system that detects fraudulent transactions and
authenticates credit card users, leveraging machine learning algorithms and diverse authentication
methods to mitigate risks and increase the security of credit card transactions.
iii
TABLE OF CONTENTS
Declaration…………………………………………………………………… i
Acknowledgement……………………………………………………............ ii
Abstract………………………………………………………………………. iii
Table Of Contents…………………………………………………………… iv
List Of Figures………………………………………………………………. vi
List Of Tables………………………………………………………………... vii
CHAPTER 1 – INTRODUCTION………………………………………………. 6
1.1 Motivation……………………………………………………………………... 8
1.2 Existing System………………………………………………………………... 9
1.3 Proposed System……………………………………………………………… 10
1.4 Objectives……………………………………………………………………… 11
1.5 Features with Scope…………………………………………………………… 12
1.6 Limitations……………………………………………………………………... 13
1.7 Organization of Report………………………………………………………... 14
CHAPTER 2 – LITERATURE SURVEY………………………………………. 15

2.1 General working features of the existing systems…………………………… 15
2.2 Different types…………………………………………………………………. 16
2.3 Table……………………………………………………………………………. 20
2.4 Technological Issues…………………………………………………………… 22
CHAPTER 3 – SYSTEM REQUIREMENT SPECIFICATION……………… 24

3.1 Analysis / Feasibility Study…………………………………………………... 25
3.2 Hardware Requirements Specification……………………………………… 26
3.3 Software Requirements Specification………………………………………… 26
CHAPTER 4 – SYSTEM DESIGN ……...………………………………………. 29

4.1 Architectural Representations………………………………………………… 29
iv
CHAPTER 5 – IMPLEMENTATION AND TESTING…...…………………… 33
5.1 General Implementation Discussions………………………………………… 34
5.2 Test Setup…………..…………………………………………………………... 34
5.3 Test Cases…………..…………………………………………………………... 36
CHAPTER 6 – RESULTS AND DISCUSSIONS…...………………………….. 38

6.1 Screenshots of the output…………..…………………………………………. 41
CHAPTER 7 – CONCLUSION……………………...…………………………… 49
7.1 Scope for Future Enhancement ……………………………………………… 49
REFERENCES……………………...……………………………………………... 51
v
LIST OF FIGURES
Fig No Fig Name Page No.

4.1 Flowchart of Proposed model 30
4.2 Block diagram 31
6.1 Heatmap 42
6.2 User Interface with indicators 44
6.3 User Interface with criteria to be filled 44
6.4 OTP Authentication level 45
6.5 Failure of face authentication 46
6.6 Successful Transaction 47
6.7 Failure of OTP Authentication 47
6.8 OTP received by User 48
6.9 BOTFather-Telegram Bot 49
6.10 Face Recognition Unknown 51
vi
LIST OF TABLES
Table No. Table Name Page. No.

Table 2.1 Literature Survey 20
Table of Test Cases 37
Table 5.3
vii
Credit Card Fraud Detection and Authentication using ML
CHAPTER 1
INTRODUCTION
1. Introduction
When coming to the issuance of the card, the condition involved is that the cardholder will
pay back the original amount they borrowed along with the additional charges they agreed to pay
Fraud can be defined as criminal deception with the intent of acquiring financial gain. The number
of online transactions has grown in large quantities and online credit card transactions hold a huge
share of these transactions. Credit card fraud events take place frequently and then result in huge
financial losses. Credit cards remain a top vector for identity theft. Credit card fraud was a close
second in the number of reported identity theft cases in 2021. In some modes of online payments,
only card numbers, expiration dates, and CVV are required and that data may be lost without our
presence, in some cases we don't even know our data is being stolen.
Credit card fraud (CCF) is a type of identity theft in which someone other than the owner
makes an unlawful transaction using a credit card or account details. A credit card that has been
stolen, lost, or counterfeited might result in fraud. Card-not-present fraud, or the use of your credit
card number in e-commerce transactions has also become increasingly common as a resultof the
increase in online shopping. Increased fraud, such as CCF, has resulted from the expansionof e-
banking and several online payment environments, resulting in annual losses of billions of dollars.
In this era of digital payments, CCF detection has become one of the most important goals. As a
business owner, it cannot be disputed that the future is heading towards a cashless culture. As a
result, typical payment methods will no longer be used in the future, and thereforethey will not be
helpful for expanding a business. Customers will not always visit the business with cash in their
pockets. They are now placing a premium on debit and credit card payments. As a result,
companies will need to update their environment to ensure that they can take all typesof payments.
In the next years, this situation is expected to become much more severe. In 2020,there were
393,207 cases of Credit Card Fraud out of approximately 1.4 million total reports of identity theft.
CCF is now the second most prevalent sort of identity theft recorded as of this year, only following
government documents and benefits fraud.
In 2020, there were 365,597 incidences of fraud perpetrated using new credit card accounts.
The number of identity theft complaints has climbed by 113% from 2019 to 2020, with credit
card identity theft reports increasing by 44.6%. Payment card theft cost the global economy
$24.26 billion last year. With 38.6% of reported card fraud losses in 2018, the United States is
Dept of CSE, SJBIT 2023-24 Page 6

the most

vulnerable country to credit theft. As a result, financial institutions should prioritize equipping
themselves with an automated fraud detection system. The goal of supervised CCF detection is to
create a machine learning (ML) model based on existing transactional credit cardpayment data.
The model should distinguish between fraudulent and nonfraudulent transactions, and use this
information to decide whether an incoming transaction is fraudulent or not. The issue involves a
variety of fundamental problems, including the system’s quick reaction time, cost sensitivity, and
feature pre-processing.
ML is a field of artificial intelligence that uses a computer to make predictions based on prior
data trends ML models have been used in many studies to solve numerous challenges.
Deeplearning (DL) algorithms applied applications in computer network, intrusion detection,
banking,insurance, mobile cellular networks, health care fraud detection, medical and malware
detection,detection for video surveillance, location tracking, Android malware detection, home
automation, and heart disease prediction. We explore the practical application of ML,
particularly DL algorithms, to identify credit card thefts in the banking industry in this paper. For
data categorization challenges, the support vector machine (SVM) is a supervised ML technique.
It is employed in a variety of domains, including image recognition, credit rating, andpublic
safety. SVM can tackle linear and nonlinear binary classification problems, and it finds
ahyperplane that separates the input data in the support vector, which is superior to other
classifiers. Neural networks were the first method used to identify credit card theft in the past .
As a result, (DL), a branch of ML, is currently focused on DL approaches. In recent years,
deeplearning approaches have received significant attention due to substantial and promising
outcomes in various applications, such as computer vision, natural language processing, and
voice. However, onlya studyhas examined the application of deep neural networks in
identifyingCCF. It uses a number of deep learning algorithms for detecting CCF. However, in
this study, we choose the CNN model and its layers to determine if the original fraud is the
normal transaction of qualified datasets. Some transactions are common in datasets that have
been labelled fraudulent and demonstrate questionable transaction behavior. As a result, we
focus onsupervised and unsupervised learning in this research paper. The class imbalance is the
problemin ML where the total number of a class of data (positive) is far less than the total
number of another class of data (negative). The classification challenge of the unbalanced dataset
has been the subject of several studies. An extensive collection of studies can provide several
answers.
We propose to alter the DL algorithm of the CNN model by adding the additional layers for
features extraction and the classification of credit card transactions as fraudulent or otherwise.
The top attributes from the prepared dataset are ranked using feature selection techniques.
Afterthat, CCF is classified using several supervised machine-driven and deep learning models.
In this study,

the main aim is to detect fraudulent transactions using credit cards with the help of ML
algorithms and deep learning algorithms. This study makes the following contributions.
Feature selection algorithms are used to rank the top features from the CCF transaction
dataset, which help in class label predictions. The deep learning model is proposed by adding a
number of additional layers that are then used to extract the features and classification from the
credit card farad detection dataset. To analyze the performance CNN model, apply different
architecture of CNN layers. To perform a comparative analysis between ML with DL algorithms
and proposed CNN with baseline model, the results prove that the proposed approach
outperforms existing approaches. To assess the accuracy of the classifiers, performance
evaluation measures, accuracy, precision, and recall are used. Experiments are performed on
thelatest credit cards dataset. The rest of the paper is structured as follows: The second section
examines the related works. The proposed model and its methodology. The dataset and
evaluation measures are described in Section 4. It also shows the outcomes of our tests on a
realdataset, as well as the analysis.
1.1 Motivation
Credit card fraud is a pervasive problem that affects both financial institutions and
customers. To combat this issue, many credit card companies are turning to machine learning
techniques for fraud detection. By utilizing machine learning, financial institutions can prevent
fraudulent transactions, minimize financial losses, and protect customers from the consequences
of identity theft.One of the primary drivers for employing machine learning in credit card fraud
detection is the sheer volume of transactional data that credit card companies process on a daily
basis. Human analysts would find it challenging to sift through and identify fraudulent activity.
Machine learning algorithms can quickly analyze large datasets and identify patterns that may
indicate fraudulent activity. By processing vast amounts of transactional data, machine learning
algorithms can detect patterns of fraudulent activity that would not be immediately apparent to
human analysts.
Fraudsters are continually altering their methods to evade detection, which makes it
challenging for traditional rule-based systems to keep pace. Machine learning algorithms can
adapt to these changing patterns and detect new types of fraud. By continually analyzing
transactional data and looking for patterns, machine learning algorithms can quickly adapt to
new types of fraud, minimizing losses due to fraudulent activity. Real-time detection of
fraudulent transactions is crucial in minimizing losses. Machine learning algorithms can
scrutinize
transactions as they occur and immediately flag potential fraud, enabling banks to take prompt
action. By analyzing transactional data in real-time, machine learning algorithms can identify
fraudulent activity before it causes significant financial losses.
Traditional fraud detection systems are prone to producing false positives, flagging
legitimate transactions as fraudulent. Machine learning algorithms can enhance accuracy by
recognizing authentic patterns of fraudulent activity and reducing false positives. By analyzing
large datasets and identifying patterns of fraudulent activity, machine learning algorithms can
accurately detect fraudulent activity while minimizing the number of false positives.
The use of machine learning techniques in credit card fraud detection can enhance
accuracy, hasten detection, and enable financial institutions to keep abreast of evolving fraud
patterns. Machine learning has emerged as a valuable tool for addressing intricate challenges
and making predictions across various domains.
1.2 Existing System
Credit card fraud detection is an important area of focus for financial institutions and
customers alike. Various algorithms and strategies have been developed for carrying out credit
card fraud detection, such as the CCFD system, which uses a Markov model, Cost Sensitive
Decision Tree (CSDT), Support Vector Machine (SVM), and Random Forest. Neural networks
have also been suggested for this purpose, and the current approach involves the use of the whale
swarm optimization technique to determine an incentive value, which is then corrected using the
BP network. However, these methods have significant drawbacks, including declining accuracy
levels, inefficiency, and challenges encountered by the cardholder during the investigation
process. Maintaining a comprehensive record of all transactions in a log is essential, but it results
in the accumulation of substantial data. With the increasing prevalence of online transactions,
identifying the cardholder is often impossible, and only the IP address is recorded for verification
purposes, necessitating the involvement of cybercrime for fraud investigations. The primary
obstacle with the existing credit card fraud detection system is the limited availability of high-
quality datasets, which are often imbalanced and contain substantial amounts of unknown
information.
To address these limitations, we propose a novel credit card fraud detection system that
utilizes the whale algorithm and SMOTE technique to enhance accuracy and efficiency while
minimizing the misclassification of normal transactions as fraudulent. The whale algorithm is a

type of swarm intelligence algorithm that is based on the behavior of humpback whales, and it has
been

shown to be effective in solving complex optimization problems. The SMOTE technique is used
to address the issue of imbalanced datasets by oversampling the minority class. The proposed
system aims to improve the accuracy and efficiency of credit card fraud detection while reducing
the number of false positives and false negatives. By using the whale algorithm and SMOTE
technique, the proposed system can adapt to changing fraud patterns and detect new types of
fraud, thereby enabling financial institutions to stay ahead of fraudulent activity.
Credit card fraud detection is a crucial aspect of financial security, and there are several
algorithms and strategies available for carrying out this task. However, these methods have
significant limitations, including declining accuracy levels and challenges encountered by
cardholders during the investigation process. It is expected that the proposed system will be useful
for financial institutions and customers alike in detecting and preventing credit card fraud.
1.3 Proposed System
The goal of this project is to implement the effective machine learning algorithms for detecting
four types of credit card fraud. The study involves a comprehensive comparison of various
machine learning approaches, utilizing a performance metric that efficiently identifies fraudulent
transactions. The dataset used in this study is analyzed using two primary techniques:
categorical analysis and numerical analysis.
To begin, the initial data in the dataset is categorical, and data cleaning and other
fundamental preparation techniques are used to prepare the raw data. The categorical data is then
converted into numerical data, and appropriate methodologies are employed for evaluation.
Machine learning algorithms are then used to determine the best algorithm using the
numerical data.In addition to traditional machine learning approaches, a new method of
detecting fraud is proposed.. This method takes into account the behaviors and patterns of
individual customers and utilizes them to identify anomalies that may indicate fraud. To
ensure the accuracy of the results, the performance of each algorithm is evaluated using a
robust performance metric. The performance metric considers various factors, such as the
number of false positives and false negatives, the overall accuracy, and the time taken to
process the data.
Overall, the study aims to identify the most effective machine learning algorithms for
detecting credit card fraud, utilizing both traditional and novel approaches. The analysis
considers both categorical and numerical data and employs a comprehensive performance metric
to evaluate the accuracy and efficiency of each algorithm. The proposed approach based on

client behavior is a unique and innovative way to detect fraud, which takes into account the
unique patterns and

behaviors of individual customers.

The aim is to create a system that will detect and authenticate credit card fraud by reviewing
transaction data, with the goal of looking at previous customer transaction information and
identifying behavioural trends. The groupings are later trained individually using various
classifiers. Then, one of the best strategies for predicting frauds can be selected based on the
classifier with the higher rating score. Accuracy, sensitivity, specificity, and precision are used
to assess how well the approaches perform.
1.4 Objectives
Developing a fraud detection system and implementing a reliable authentication system are
essential to ensure the security and reliability of credit card transactions. By training the system on a
dataset of past fraudulent activities, analyzing transaction data in real-time, and updating the system
to keep up with technological advancements, financial institutions can prevent fraudulent activities
and ensure the safety of consumers' financial transactions. The specific objective of this project was-
 Developing a fraud detection system that can accurately identify and classify fraudulent and
non-fraudulent credit card transactions using machine learning techniques is essential to
ensure the security and reliability of financial transactions. With the increasing number of
credit card transactions, it is crucial to prevent fraudulent activities, which can lead to
significant losses for both consumers and financial institutions. A machine learning-based
fraud detection system can analyze a vast amount of transaction data to identify patterns
and anomalies in real-time,
thereby detecting fraudulent activities and preventing them before they occur. This system can
also classify transactions into fraudulent or non-fraudulent categories, increasing its accuracy
over time.
 Implementing a reliable authentication system is equally important to verify the legitimacy of
credit card users. This system should use various authentication methods, such as biometric
authentication, one-time passwords, and security questions, to ensure the identity of the
cardholder. Biometric authentication involves using fingerprint or facial recognition to
authenticate the user, while one-time passwords are temporary passwords generated for each
transaction. Security questions are also used to verify the user's identity. A reliable
authentication system can prevent unauthorized access to credit card information and ensure
the security of transactions.
 Training the system on a dataset of past fraudulent activities is critical to identifying patterns
and recognizing similar fraudulent transactions. This dataset should contain information about
past fraudulent and non-fraudulent transactions, including transaction amount, location, time,
and the cardholder's information. By analyzing this dataset, the system can learn to identify
patterns and anomalies in credit card transactions and increase the accuracy of the fraud
detection algorithm. This training process is ongoing, and the system should continually learn
from new data to improve its accuracy over time.
 Increasing the security of credit card transactions by detecting and preventing fraudulent
activities before they occur is essential to ensuring the reliability of financial transactions. By
analyzing transaction data in real-time, the system can detect and prevent fraudulent activities,
leading to significant savings for consumers and financial institutions. This system should be
able to identify suspicious transactions and flag them for further investigation, preventing
fraudulent activities from occurring.
 Keeping up with the advancements in technology and adapting the system to new fraudulent
activities and authentication methods is essential to maintain its effectiveness. Fraudulent
activities are continually evolving, and fraudsters are always finding new ways to bypass
security systems.The fraud detection and authentication system should be updated regularly to
keep up with the latest advancements in technology and ensure its effectiveness. This includes
updating the machine learning model, incorporating new authentication methods, and
analyzing new datasets to identify patterns in fraudulent activities.
1.5 Features with Scope
Credit card fraud has been a persistent problem for both financial institutions and customers.
To mitigate financial losses for banks and safeguard customers against identity theft, machine
learning techniques are increasingly being utilized in credit card fraud detection. There are
several factors to consider when developing a credit card fraud detection system, and the
following are three crucial drivers that should be considered in any approach.
• Early Warning. - A credit card fraud detection system should identify fraudulent
activity swiftly and accurately. Fraudulent transactions can cause significant financial
losses, and it is essential to detect them as early as possible to mitigate their impact.
Machine learning algorithms can swiftly analyze large datasets and identify patterns
that
may indicate fraudulent activity. This early warning system can help banks take
prompt action to prevent further fraudulent activity.
• Fraud Elimination - A legitimate transaction should not be labeled as fraudulent.
This is important to prevent false positives, which can be detrimental to a customer's
financial standing and lead to mistrust in the financial institution.
The credit card fraud detection system should be designed to identify only those
transactions that are genuinely fraudulent. Prior to every major event or incident,
such as holiday shopping, the system should be able to identify scams that are about
to occur and prevent them from happening.
• Simplicity of Implementation - The algorithm used in the credit card fraud
detection system should be simple to implement since hardware and software
requirements are easily met. The system should be user-friendly and not require
extensive technical expertise to use. This will ensure that the system can be easily
integrated into the existing infrastructure of financial institutions, making it a cost-
effective solution.
The use of machine learning techniques in credit card fraud detection has the
potential to prevent fraudulent transactions, mitigate financial losses for banks, and
safeguard customers against the repercussions of identity theft. To develop an
effective credit card fraud detection system, it is essential to consider the drivers of
early warning, fraud elimination, and simplicity of implementation. By taking
these drivers into account, banks can implement a system that accurately identifies
fraudulent activity while minimizing false positives, is easy to use.
1.6 Limitations
Credit card fraud detection and authentication systems that use machine learning algorithms
have certain limitations that must be considered.
 One of the limitations is a high false positive rate in the system, which can cause valid
transactions to be rejected. This can lead to customer dissatisfaction and may discourage them
from using the service in the future. Additionally, a high false negative rate can result in the
approval of fraudulent transactions, which could result in significant financial losses for both the
customers and the financial institutions
 Another limitation is that the algorithm may not be able to identify new or previously
unknown types of fraud, making it difficult to keep up with emerging fraudulent activities.
Furthermore, the

system may not be able to detect complex fraud schemes that are designed to evade detection.
 It is also possible that the system may not be able to process a large volume of transactions
in real-time, leading to delays and potential system failures. Finally, the implementation of such
systems can be expensive, as they may require a significant amount of computational power,
which can increase the cost of implementation.
While machine learning algorithms have proven to be effective in detecting credit card fraud
and providing authentication services, it is important to consider the limitations of the systems
and implement measures to mitigate these issues.
1.7 Organization of report
In order to make it easier to understand and organize, this report has been given in
seven chapters. Introduction, Literature Review, System Requirement Specification, System
Design, Implementation & Testing, Discussion Results, and Conclusion are the order in which
they should be presented.

CHAPTER 2
LITERATURE SURVEY
2. 1 General working features of the existing system
• Data Gathering: The system gathers information about transactions from a range of sources,
including as point-of-sale terminals, internet transactions, and mobile apps.
• Data pre-processing: To get rid of outliers, inconsistencies, and missing data, the system pre-
processes the data. In order to get the data ready for more analysis, this stage also comprises
normalising and scaling it.
• Feature Extraction: The system takes the preprocessed data and extracts pertinent features such
transaction amount, location, time, and device details.
• Rule-based Analysis: To identify possibly fraudulent transactions, the system applies a set of
predetermined criteria to the retrieved features. These regulations frequently draw from past
information and standard operating procedures.
• Machine Learning Analysis: The system use machine learning algorithms to identify fraudulent
transactions and understand patterns and relationships in the data. These algorithms might use
unsupervised learning strategies like clustering and anomaly detection or supervised learning
strategies like logistic regression, decision trees, and neural networks.
• Risk Scoring: Using the findings of rule-based and machine learning analysis, the system rates the
risk of each transaction. Low-risk transactions can continue while high-risk transactions are
highlighted for additional examination.
• Making a decision: The system decides whether to allow or reject a transaction based on the risk
scores as well as other elements including client history and account information.
• Continuous Learning: The system continuously adjusts its models and rules based on fresh data in
order to increase accuracy and efficacy over time.

2.2 Different Types

Parth Roy, Prateek Rao, Jay Gajre [1]. The International Fraud Awareness organization (IFA)
has proposed the use of machine learning techniques to detect fraudulent Master Card transactions.
This approach aims to enhance the current fraud detection system by reducing false alarm rates and
improving the detection rate of scams. With data on card transactions continuing to be collected, the
methods can be modified and implemented in the bank's credit card fraud detection system to identify
and prevent fraud. As European cardholders have reported 284,807 communications related to fraud,
the need for an effective fraud detection system is crucial to safeguard financial institutions and
customers alike. The utilization of machine learning techniques can analyze vast amounts of
transactional data and detect patterns that may indicate fraudulent activity. By identifying and
stopping fraudulent transactions promptly, banks can mitigate financial losses while safeguarding
customers against identity theft. Overall, implementing machine learning in the credit card fraud
detection system is an important step towards addressing the intricate challenges posed by fraudulent
activity
Ishika Sharma, Shivjyoti Dalai, Venktesh Tiwari, Ishwari Singh , Seema Kharb [2] presented
the problem of credit card fraud detection can be addressed using various techniques such as Naive
Bayes, Random Forest, and Logistic Regression. Each transaction is evaluated individually, and the
most effective method is used for fraud detection. The main objective is to identify and prevent fraud
by filtering the available strategies to achieve the best possible outcome. These techniques are applied
to the credit card data to develop a fraud detection system that accurately detects fraudulent
transactions while minimizing false positives. The selected method must be efficient and effective in
detecting fraud in real-time while minimizing the impact on legitimate transactions. Overall, the goal
of these techniques is to improve the accuracy and efficiency of credit card fraud detection, ensuring
that legitimate transactions are not labeled as fraudulent and that fraudulent transactions are quickly
identified and stopped.
Anuruddha Thennakoon, Chee Bhagyani, Sasitha Premadasa, Shalitha Mihiranga, Nuwan
Kuruwitaarachchi [3] offered an evaluation that offers a thorough manual for choosing the best
algorithm for the kind of frauds, and we use an adequate performance metric to show the evaluation.
In order to determine if a particular transaction is legitimate or fraudulent, they also created the use of
predictive analytics performed by the implemented machine learning models and an API module.
H. Najadat, O. Altiti, A. A. Aqouleh, and M. Younes [4] conducted an experimental study to
address the issue of imbalance classification and evaluated machine learning methods for fraud
detection. They examined various options and identified the weaknesses in current machine learning

fraud detection methods. Using a dataset labelled with credit card fraud, they summarized their

findings. The study involved a comprehensive investigation of the problem, highlighting potential
areas of improvement in current machine learning approaches to combat fraud detection. The
researchers sought to improve the accuracy of fraud detection while reducing false positive rates.
Their analysis has the potential to inform and improve fraud detection systems, particularly in the
financial sector, where fraudulent transactions can have significant consequences. The study is a
significant contribution to the field of machine learning for fraud detection, providing valuable
insights into how to approach this complex problem.
S. Makki, Z. Assaghir, Y. Taher, R. Haque, M.-S. Hacid, and H. Zeineddine [5] presented
several machine learning-based classification techniques, such as Naive Bayes, Random Forest, and
Logistic Regression, to address the issue of imbalanced datasets. These techniques were evaluated
using various performance metrics, including accuracy, precision, recall, f1 score, confusion matrix,
and Roc-auc score. The main objective was to identify the best approach for detecting credit card
fraud by applying these techniques to the highly imbalanced dataset. The results of the study showed
that Random Forest outperformed the other techniques in terms of accuracy, precision, recall, f1
score, and Roc-auc score. This finding suggests that Random Forest is a reliable technique for fraud
detection tasks, and its performance can be further improved by adjusting its parameters. Overall, the
study provides valuable insights into the use of machine learning techniques for fraud detection and
emphasizes the importance of selecting appropriate metrics for evaluating their performance.
D. Tanouz, R Raja Subramanian, D. Eswar, G V Parameswara Reddy, A. Ranjith kumar, CH
V N M Praneeth[6] proposed using several machine learning algorithms, including Random Forest,
Gradient Boost, and Support Vector Machine, individually and in combination, depending on the data
set and application. Despite performing multiple computations, they found that all algorithms
experienced some level of imbalance during the study. They plotted learning curves and observed
that while logistic regression had higher accuracy, KNN was only able to learn, and the majority of
the algorithm was underfit. Thus, KNN was deemed a stronger classifier for detecting credit card
fraud. The researchers from the university along with students and also working on the same project
and laid emphasize on the need for ongoing evaluation and improvement of machine learning models
for fraud detection to keep up with evolving fraudulent tactics.
MJ Madhury, H L Gururaj, B C Soundarya, K P Vidyashree ,B Rajendra [7] provided a
variety of machine learning-based methods for credit card recognition, including XG Boost, Decision
Tree, Random Forest, Support Vector Machine, and Extreme Learning Method. To get effective
results, comparative examination of both machine learning and deep learning algorithms was done.
Using the European card benchmark dataset for fraud detection, a thorough empirical investigation is
conducted. The dataset was first subjected to a machine learning technique, which somewhat
increased the

accuracy of fraud detection.The evaluation of the research effort demonstrates the enhanced
outcomes obtained, with optimised values for accuracy, f1-score, precision, and AUC Curves of
99.9%, 85.71%, 93%, and 98%, respectively. For situations involving credit card detection, the
suggested model performs better than cutting-edge machine learning and deep learning techniques.
Fawaz Khaled Alarfaj , Iqra Malik , Hikmat Ullah Khan , Naif Almusallam , Muhammad
Ramzan , And Muzamil Ahmed [8] proposed machine learning models were created for payment
card fraud detection using both publicly available and actual transaction records. The performance of
these models was analyzed and compared using the original features and the aggregated features. To
evaluate the effectiveness of using aggregated features, a statistical hypothesis test was conducted to
determine if they have greater discriminative ability than the original features in detecting fraud. The
findings indicate that utilizing aggregated features is an effective approach to address real-world
payment card fraud detection issues. The study suggests that the use of a genetic algorithm to create
combined characteristics could improve the models' performance in detecting fraud. By comparing
the results of the 13 models, it was concluded that the use of a Gradient Boosting Decision Tree
algorithm with aggregated features was the most effective in identifying fraudulent transactions. The
study demonstrates the importance of combining different techniques and features to improve the
accuracy of payment card fraud detection.
Manjeevan Seera, Chee Peng Lim, Ajay Kumar, Lalitha Dhamotharan, Kim Hua Tan [9]
proposed a method to determine whether transactions on the Kaggel-provided IEEE-CIS Fraud
Detection dataset were genuine or fraudulent. Bidirectional Long Short-Term Memory (BiLSTM)
and bidirectional Gated Recurrent Unit (BiGRU) are the foundations of the model, which is called
BiLSTM- MaxPooling-BiGRUMaxPooling. They used the following six machine learning
classifiers: Logistic Regression, Naive Base, Voting, Ada Boosting, Random Forest, and Decision
Tree. When comparing the outcomes of machine learning classifiers and our model, it is clear that the
model performed better because it received a 91.37% score.
Sailusha Ruttala; Gnaneswar V.; Ramesh R.; Rao, G. Ramakoteswara [10] the adaboost and
random forest algorithms are commonly used in fraud detection. In this context, the researchers
evaluated the performance of these two algorithms based on various metrics such as F1-score,
accuracy, precision, recall, and other performance measures. The confusion matrix and ROC curve
were used to analyze the outcomes of these algorithms. The researchers compared the results of both
algorithms and determined the algorithm with the highest accuracy, precision, recall, and F1-score to
be the best one for detecting fraud. This evaluation helps in determining which algorithm to use for
fraud detection based on specific needs and circumstances. The findings can also help improve the
accuracy and efficiency of fraud detection systems, thus saving organizations and individuals from
financial loss.

Overall, this research contributes to the development and enhancement of fraud detection
technologies and helps in combating fraud in various industries, including finance, healthcare, and e-
commerce
Xuan Shiyang Xuan,Guanjun Liu, Zhenchuan Li,Shuo Wang, Lutao Zheng, Changjun Jiang [11]
use of two different random forest models for training typical and anomalous transaction behaviours
and comparing their effectiveness in detecting credit card fraud. The two random forests use different
underlying classifiers and are compared based on their performance metrics. The study found that
both random forest models effectively detected credit fraud, but the model with the linear
discriminant analysis classifier performed slightly better in terms of precision, recall, and F1-score.
The article also discusses the importance of selecting appropriate features for training the random
forests and highlights the significance of behavioural features in improving the fraud detection
accuracy. Overall, the study provides insights into the use of random forest models in credit card
fraud detection and the importance of considering different classifiers for optimal performance.
S. Makki, Z. Assaghir, Y. Taher, R. Haque, M.-S. Hacid, and H. Zeineddine [5] presented
several machine learning-based classification techniques, such as Naive Bayes, Random Forest, and
Logistic Regression, to address the issue of imbalanced datasets. These techniques were evaluated
using various performance metrics, including accuracy, precision, recall, f1 score, confusion matrix,
and Roc-auc score. The main objective was to identify the best approach for detecting credit card
fraud by applying these techniques to the highly imbalanced dataset. The results of the study showed
that Random Forest outperformed the other techniques in terms of accuracy, precision, recall, f1
score, and Roc-auc score. This finding suggests that Random Forest is a reliable technique for fraud
detection tasks, and its performance can be further improved by adjusting its parameters. Overall, the
study provides valuable insights into the use of machine learning techniques for fraud detection and
emphasizes the importance of selecting appropriate metrics for evaluating their performance.
D. Tanouz, R Raja Subramanian, D. Eswar, G V Parameswara Reddy, A. Ranjith kumar, CH V
N M Praneeth[6] proposed using several machine learning algorithms, including Random Forest,
Gradient Boost, and Support Vector Machine, individually and in combination, depending on the data
set and application. Despite performing multiple computations, they found that all algorithms
experienced some level of imbalance during the study. They plotted learning curves and observed
that while logistic regression had higher accuracy, KNN was only able to learn, and the majority of
the algorithm was underfit. Thus, KNN was deemed a stronger classifier for detecting credit card
fraud. The researchers from the university along with students and also working on the same project
and laid emphasize on the need for ongoing evaluation and improvement of machine learning models
for fraud detection to keep up with evolving fraudulent tactics.
.

2.3 Related papers in table format
Table 2.3 Literature Survey
Sl. Paper Author, Publisher Problem Technologi Outcome

No Title & Publication Identified es Used
Year
1 Comprehe Parth Roy, Prateek Data is Logistic IF achieved
nsive Rao, Jay imbalanced, regression, considerable
Analysis Garjre,Kanchan low isolation better performance
for Fraud Katake,Arvind accuracy, forest, than LOF through
Detection Jagtap,Yogesh presence of Support method of we
of Credit Gajmal.Internationa inappropriate Vector might understand
Card l Conference on positives. Machine & that scam cases
through Emerging Smart Local remains about 27
Machine Computing and Outlier % against LOF
Learning Informatics Factor gratitude
(ESCI),March algorithm. proportion of fair 2
2021. %, SVM of 0%,
Logistic
Regression has
detection rate of
50%.
2 Credit Ishika Low Logistic Logistic
Card Sharma,Shivjyoti accuracy, and Regression, Regression,
Fraud Dalai, Venktesh imbalanced Random Random Forest,
Detection Tiwari, Ishwari dataset is Forest, Naïve Bayes.
Using Singh, Seema used. Hybrid Naïve
Machine Kharb,International algorithm will Bayes.
Learning Research Journal result even
& Data of Engineering and better.
Science Technology
(IRJET), June 2022
3 Real-time Anuruddha Provide an Support The developed
Credit Thennakoon, Chee average level Vector machine learning
Card Bhagyani, Sasitha of accuracy, Machine, models present an
Fraud Premadasa, focuses on Naive average level of
Detection Shalitha improving the Bayes, K- accuracy,
Using Mihiranga, Nuwan prediction Nearest
Machine Kuruwitaarachchi levels to Neighbor
Learning ,IEEE 9th acquire a and Logistic
International better Regression
Conference on prediction.
Cloud Computing, Does not
Data Science & focus on to
Engineering, 2019 focus on
location-based
frauds.

4 Credit D. Tanouz, R Raja Insufficient Random It couldn't reach

Card Subramanian, D. dataset forest, 100% accuracy as
Fraud Eswar, G V from banks, logistic the dataset used is
Detection Parameswara False regression, not sufficient and
Using Reddy, A. Ranjith Negatives , naive bayes there were False
Machine kumar, CH V N M Must train Negatives.
Learning praneeth, Fifth models with
International more data.
Conference on
Intelligent
Computing and
Control Systems
(ICICCS 2021)
5 Explorator MJ Madhury HL Mainly Naïve Bayes Though the used
y analysis Gururaj uses the algorithm,D techniques and
of credit BCSoundarya KP Naive ecision tree algorithms are
card fraud Vidyashree ,B Bayes algorithmK- proven to be
detection Rajendra ,Elsevier Algorithm NN efficient and
using B.V. 2022 algorithm, accurate while
machine which Random being implemented
learning contains a forest individually or in
techniques very low classifiers combinations,
. precision there could be a
rate. chance of having
dissimilar and
irregular datasets.
6 Credit Fawaz Khaled The Extreme Comparing all the
Card Alarfaj , Iqra performance learning algorithm
Fraud Malik on unseen method,rand performances side
Detection , Hikmat Ullah data increased om forest to side, the CNN
Using Khan , Naif as the class (rf), logistic with 20 layers and
State-of- Almusallam , imbalance regression the baseline model
the-Art Muhammad increased and is the top method
Machine Ramzan , And there is now a with an accuracy
Learning Muzamil Ahmed work based on of 99.72%.
and Deep ,This work was it.
Learning supported by the
Algorithm Deanship of
s Scientific through
the Research 2022.
7 An Manjeevan Seera, Finding out Support The best AUC

intelligent Chee Peng Lim, the best vector score (from GBT)
payment Ajay Kumar, machine machines,R is 0.866, while the
card Lalitha learning ule best AUC score
fraud Dhamotharan, Kim algorithms induction,N (also from GBT)
detection Hua Tan ,Springer with the eural increases to 0.967
system Science+Business highest networks,Tr with the use of
Media, LLC, part of accuracy rate ee methods aggregated
by a features. RF has

Springer Nature statistically recorded the

2021. done largest
comparative improvement in
study. But AUC, i.e., 0.5 with
lacks the original feature
implementatio set to 0.958 with
n. the aggregated
features.
8 Credit Hassan Najadat, The best AUC Random The results from
Card Ola Altiti , Ayah was only 80% under machine learning
Fraud Abu Aqouleh, and 81% was sampling, classifiers show
Detecti Mutaz Younes, achieved by Random that the best
on 11th International hard voting over AUC(Area under
Based Conference on with sampling, Receiver operating
on Information and undersamplin SMOTE characteristic
Machin Communication g and Curve) was 80%
e and Systems (ICICS), oversampling and 81% that
Deep April 2020 techniques. achieved by hard
Learnin voting with under
g sampling and
oversampling
technique.
9 Credit Sailusha Ruttala; The results Random Random Forest

Card Gnaneswar V.; are not Forest Algorithm works
Fraud Ramesh R.; Rao, G. satisfactory. Algorithm best than the
Detection Ramakoteswara, and Adaboost
Using International Adaboost algorithm to detect
Machine Conference on algorithm. credit card fraud.
Learning Intelligent
Computing and
Control Systems
,IEEE Xplore 2020.
10 Random Sailusha Ruttala; Random Random- This paper has
forest for Gnaneswar V.; forest obtains tree-based examined the
credit card Ramesh R.; Rao, G. good results random performance of
fraud Ramakoteswara, on small set forest. two kinds of
International
detection data, there are CART- random forest
Conference on
Intelligent
still some based models. A real-life
Computing and problems such random B2C dataset on
Control Systems as imbalanced forest credit card
,IEEE Xplore 2020 data, (Classificati transactions is
considers only on and used in
few attributes. Regression experiment.
Trees)
2.4 Technological Issues

Python is a high-level, interpreted programming language that has become popular for its
simplicity, readability, and flexibility. However, like any programming language, it also has some
technological issues that developers should be aware of. Here are some of the main technological
issues of Python:
 Performance: Python is an interpreted language, which means it runs slower than
compiled languages like C or C++. This can be a problem for performance-intensive
applications or large-scale data processing.
 Global Interpreter Lock (GIL): The GIL is a mechanism in Python that prevents
multiple threads from executing Python code in parallel. This can limit the performance of
multi- threaded applications in Python.
 Memory management: Python uses garbage collection to manage memory automatically,
but this can lead to performance issues when dealing with large amounts of data.
 Limited mobile app development support: Python is not well-suited for mobile app
development, as it is not a native language for iOS or Android development.
 Limited support for machine learning hardware: Python's standard libraries are not
optimized for specialized hardware like GPUs, which can limit performance in machine
learning applications.
 Lack of backward compatibility: Python has gone through several major version
updates, which can cause compatibility issues between different versions of
the language.
 Security concerns: Like any programming language, Python is vulnerable to security
issues like SQL injection and cross-site scripting if not used properly.

CHAPTER 3
SYSTEM REQUIREMENT SPECIFICATION

The System Requirements Specification (SRS) is a vital document that outlines the
requirements for a software or information system project. The SRS document typically includes
the following information:
 Introduction: This section provides an overview of the software project and explains the
purpose of the SRS document.
 Scope: This section defines the boundaries of the system, including what it will and will
not do.
 Functional Requirements: This section defines the specific functions that the software
must perform, including inputs, outputs, and processing requirements.
 Non-functional Requirements: This section defines the qualities that the software must
possess, including performance, reliability, usability, and security.
 Design Requirements: This section defines the software architecture, data structures, and
other design-related requirements.External Interfaces: This section defines the
requirements for external interfaces, such as user interfaces, hardware interfaces, and data
interfaces.
 Constraints: This section defines any limitations or restrictions that may impact the
development or deployment of the software.
 Assumptions and Dependencies: This section identifies any assumptions or dependencies

that must be considered during the development process.
In addition to the SRS document, a feasibility study is often conducted to determine whether the
software project is viable and to identify any potential obstacles or risks. The feasibility study
typically includes a market analysis, technical analysis, and financial analysis.
Hardware and software requirement specification is another important component of the SRS
document. This section outlines the specific hardware and software requirements
needed for the software to function properly. This includes information on the operating system,
processors, memory, storage, and any other hardware or software components needed for the
software to operate as intended.

Overall, the SRS document serves as a vital blueprint for the software development team,
ensuring that all stakeholders have a clear understanding of the project requirements and that the
software is developed in a way that meets those requirements.
3.1 Analysis / Feasibility Study
A feasibility study is an assessment of whether a proposed project or system has the potential
to meet its objectives. In the case of the proposed approach for credit card fraud detection and
authentication system using machine learning, the feasibility study would consider the following
aspects:
 Technical feasibility: This involves evaluating whether the proposed approach is

technically feasible, based on factors such as the availability of data, computing resources,
and expertise required to develop and deploy machine learning models.
 Operational feasibility: This involves evaluating whether the proposed approach is

practical and can be implemented within the existing operational environment, taking into
account factors such as integration with existing systems and the impact on business
processes.
 Economic feasibility: This involves evaluating the cost-effectiveness of the proposed

approach, taking into account factors such as the cost of acquiring and maintaining
machine learning models, the cost of false positives and false negatives, and potential cost
savings from fraud prevention.
 Schedule feasibility: This involves evaluating whether the proposed approach can be
implemented within the desired timeframe, based on factors such as the complexity of the
machine learning models and potential disruptions to existing operations.
 Legal and regulatory feasibility: This involves evaluating whether the proposed
approach complies with relevant laws, regulations, and industry standards, including data
privacy and security regulations.
 Ethical feasibility: This involves evaluating the ethical implications of using machine
learning for credit card fraud detection, including potential bias and discrimination in the
model outputs.
Overall, the feasibility study would assess the potential risks and benefits of the proposed
approach and help determine whether it is viable and worth pursuing. It would also help identify

any potential obstacles or challenges that would need to be addressed before the system can be
successfully implemented.
3.2. Hardware Requirement Specification

The hardware requirement specification (HRS) for a software project is a document that
outlines the necessary hardware components for the software to function properly. In this case, we
are considering the HRS for a proposed software project. The HRS provides a clear guideline for
the type of computer system that is required to run the software application without any
performance issues.
The HRS document specifies that the computer system should have a Pentium IV 2.4 GHz
processor, which is a relatively old processor by today's standards. However, this processor
should be sufficient to run the proposed software application. The processor's clock speed is a
critical factor in determining the system's performance, and 2.4 GHz is considered an acceptable
speed for most basic software ‘applications.
The document also specifies that the computer system should have a hard disk capacity of 500
GB. This amount of storage space is sufficient for most software applications, and it should be
more than enough for the proposed project. The hard disk capacity is an essential factor to
consider because it determines how much data the computer system can store. It is important to
have enough storage space to avoid running out of storage while using the software application.
The HRS document also specifies that the computer system should have 4 GB of RAM. RAM
(Random Access Memory) is a type of memory that is used to temporarily store data that the
computer system is currently working on. The amount of RAM available in a computer system is
a critical factor in determining its performance. 4 GB of RAM is sufficient for most basic
software applications, and it should be sufficient for the proposed project.
The HRS document outlines the necessary hardware requirements for the proposed software
project. The specifications provided in the document are not particularly high-end, but they are
sufficient to run most basic software applications. It is important to note that any desktop or
laptop system that meets or exceeds the specifications outlined in the HRS document can be used
to run the software application. It is also important to ensure that the system's hardware
components are compatible with the software application to ensure optimal performance.
3.3. Software Requirements
Software requirements are an essential part of software development. They specify the
software environment in which the software application will operate. The software requirements
include the

operating system, programming language, software, and other factors that are necessary for the
software application to function correctly. In this case, the software requirements for the project
include the operating system, coding language, software, and language.
Operating System
The operating system is the software that manages the computer's hardware resources and
provides services to the software applications. The software application in this project requires a
Windows 7 and above operating system. The choice of operating system is critical because it
affects the software application's compatibility and performance. The software application may
not function correctly if the operating system does not meet the minimum requirements specified
in the software requirements specification (SRS) document. Therefore, it is essential to ensure that
the operating system meets the software requirements.
Coding Language
The coding language is the programming language used to develop the software application.
In this project, the software application is developed using Python v 3.7
and above. Python is a popular high-level programming language used for data analysis, machine
learning, and artificial intelligence applications. Python is known for its ease of use, readability,
and versatility. It is also open source, which means that developers can modify and redistribute
the code. The choice of coding language is essential in determining the software application's
functionality, ease of use, and maintenance.
Software
The software is the set of programs and data used to support the software application. In this
project, the software application requires Anaconda Navigator. Anaconda Navigator is an open-
source software that provides an easy-to-use interface for managing packages and environments
for data science and machine learning tasks. Anaconda Navigator comes with pre-installed Python
packages and tools, making it an ideal platform for data science projects. The use of Anaconda
Navigator in the software application indicates that it may involve data analysis or machine
learning tasks.
Language
The language is the natural language used to write the software application. In this project, the
software application is developed using Python programming language. Python is a high-level,
interpreted programming language that is easy to learn and read. Python is known for its syntax,

which makes it easy to write programs quickly. The use of Python in the software application
indicates that it may involve data analysis or machine learning tasks.
The software requirements are critical to ensure that the software application runs smoothly
and as intended. The software requirements specify the operating system, coding language,
software, and language used to develop the software application. The operating system, coding
language, and software are essential in determining the software application's functionality, ease
of use, and maintenance. Therefore, it is important to ensure that the software requirements are
met to ensure that the software application runs correctly.

CHAPTER 4
SYSTEM DESIGN
4.1 Architectural Representation
The architecture for the fraud detection algorithm will involve several components, including
data collection and analysis, machine learning-based classification algorithms, and a decision-
making mechanism. The architecture may include the following components:
 Data Collection: This component will collect transactional data from credit card
companies, including cardholder information, transaction history, transaction location, and
transaction amount.
 Data Analysis: This component will analyze the collected data to identify patterns and
anomalies in spending behavior that may indicate fraudulent activity. The analysis may use
statistical methods, machine learning algorithms, or a combination of both.
 Machine Learning-Based Classification Algorithms: This component will use machine
learning-based classification algorithms to classify transactions as either fraudulent or non-
fraudulent based on the patterns and anomalies identified in the data analysis phase. The
algorithms may include decision trees, random forests, support vector machines, or neural
networks.
 Decision-Making Mechanism: This component will use the output from the machine
learning-based classification algorithms to make a decision on whether a transaction is
fraudulent or non-fraudulent.
The decision-making mechanism may include a rule-based system or a more sophisticated
approach, such as a reinforcement learning algorithm. Overall, the architecture will involve a
feedback loop, where the output from the decision-making mechanism is used to improve the
accuracy of the machine learning-based classification algorithms in future transactions. The
architecture will need to be designed to handle large volumes of data and process transactions in
real-time to ensure efficient and effective fraud detection.
The flowchart illustrates the process of credit card fraud detection and authentication
using machine learning and a two-way authentication approach with OTP and face detection. It
begins with receiving transaction data, followed by pre-processing and rule-based checks to identify
suspicious patterns. An ML model analyses the data, assigning risk scores to transactions. Threshold
checks compare the risk scores against predefined thresholds. For flagged transactions, OTP
verification is initiated, and simultaneously, face detection captures the cardholder's face. Face

recognition algorithms verify the cardholder's identity. Based on successful verification, a decision is
made regarding the transaction. Reporting provides insights into fraud patterns and system
performance. The flowchart concludes, representing the termination of the system.
Fig 4.1 Flow Chart of the proposed model
The above diagram gives the representation of the flow of the project. The client database
which is accessed from the dataset and are classified as legal pattern and fraudulent pattern which
are then trained using machine learning algorithm and the test data which are the new transactions
are tested using the algorithms. In the processing stage of the transaction after its identified as
fraud
, before the transaction proceeds we have two authentication steps that occurs. Firstly we have an
OTP(One Time Password) sent to the customers phone number which is linked to the account.

Once the OTP is verified, we have a face recognition authentication.. Based on the verification i t
is the

processed or cancelled. Thereby preventing the fraudulent transactions.
Fig 4.2 Block Diagram of the proejct
The architecture diagram represents a credit card fraud detection system using machine
learning. The process starts with the Credit Card System, which generates transaction data. This data
is then passed through the Data Processing component, where it is cleaned and prepared for analysis.
The next step is Feature Engineering, where relevant features are extracted from the processed data,
such as transaction amounts, locations, and timestamps. The
Machine Learning component takes the engineered features and trains a model using machine
learning algorithms. This model learns patterns and characteristics of fraudulent transactions based on
historical data. Once the model is trained, the Fraud Detection component utilizes it to identify
potentially fraudulent transactions. If a transaction is flagged as suspicious, an alert is raised for
further investigation by relevant parties. Finally, the Alert/Reporting component is responsible for
notifying the appropriate individuals or teams about the potential fraud cases. It may also generate
reports on fraud activity and patterns, providing valuable insights for ongoing fraud prevention
efforts. Overall, this architecture enables the automated detection of credit card fraud by leveraging
machine learning techniques, helping to protect cardholders and financial institutions from fraudulent
activities.
Credit card fraud is a common problem faced by credit card companies, financial

institutions, and cardholders. To combat this problem, credit card companies have developed
advanced fraud

detection systems that leverage machine learning algorithms and statistical techniques to detect
fraudulent transactions. Here is a detailed explanation of how such a system works:
Transactional Data Collection: The credit card company sends transactional data to the fraud
detection system. This data contains information about the transaction, such as the amount,
location, time, and type of transaction.
Preprocessing: The fraud detection system collects the transactional data and preprocesses it for
analysis. Preprocessing involves cleaning and transforming the data so that it can be analyzed
effectively.
Data Analysis: The data analysis component analyzes the transactional data using statistical
methods and machine learning algorithms to identify patterns and anomalies in spending
behavior. Machine learning algorithms can learn from historical data to identify patterns and
trends that are indicative of fraudulent activities.
Classification: The machine learning-based classification algorithms component classifies

transactions as either fraudulent or non-fraudulent based on the patterns and anomalies identified
in the data analysis phase. This component can identify previously unknown fraudulent activities
and adapt to new fraud techniques.
Decision-making: The decision-making mechanism component makes a decision on
whether a transaction is fraudulent or non-fraudulent based on the output from the machine
learning- based classification algorithms. This component uses rules and thresholds set by the
credit card company to determine the likelihood of fraud and whether to alert the cardholder and
the credit card company.

CHAPTER 5
IMPLEMNTATION AND TESTING
Implementation and testing are important stages in the development and deployment of any
solution, including optimization algorithms. Here are some reasons why implementation and
testing are crucial:
 Verification of Algorithm: Converting a theoretical idea into usable computer code or

software is the first step in putting an optimization method into practice. It is possible to
confirm that an algorithm has been appropriately implemented and is operating as
expected by testing the implementation. It guarantees that the desired operations are being
carried out correctly by the algorithm and that the outcomes attained are in line with those
anticipated.
 Identification of Bugs and Errors: The software's faults or errors can be found and
rectified during implementation and testing. The effectiveness and dependability of the
solution can be significantly impacted by bugs or faults since they can produce inaccurate
results or cause the algorithm to behave in unanticipated ways. Thorough testing enables
prompt debugging and correction by assisting in the discovery of such problems.
 Performance Evaluation: It is possible to assess the performance of the optimisation

algorithm by implementation and testing. Benchmarking and comparisons with other
current approaches or standards are possible. The effectiveness, precision, and efficiency
of the algorithm in resolving the current issue are all evaluated through performance
analysis. Furthermore, it aids in locating potential zones for development and
improvement.
 Validation of Results: The findings obtained are validated by testing the implemented
method. It guarantees that the solution is generating significant, dependable results that
correspond to the anticipated consequences. In order to determine whether a solution is
credible, reliable, and appropriate for use in practical applications, validation is crucial.
 Robustness and Scalability Testing: Evaluation of the solution's resilience and

scalability can also be done during implementation and testing. Testing for robustness is
putting the solution through a variety of scenarios and settings to make sure it can handle a
variety of inputs and situations without breaking or giving inaccurate results. In summary,
implementation and testing are important stages in the development and deployment of

optimization algorithms as they help verify the correctness of the implementation, identify
and fix bugs or errors, evaluate performance, validate results, and assess robustness and

scalability. These stages ensure that the solution is reliable, effective, and suitable for real-
world applications.
5.1 General Implementation Discussions
1. Collect transactional data: Gather a large amount of transactional data from various
sources, such as credit card companies, banks, and financial institutions.
2. Preprocess the data: Clean and preprocess the data by removing any inconsistencies or
errors and organizing it into a structured format for analysis.
3. Train machine learning algorithms: Train machine learning algorithms, such as decision
trees and random forests, on the transactional data to recognize patterns and anomalies that
are indicative of fraudulent activity.
4. Incorporate diverse authentication methods: Implement diverse authentication methods,

such as facial detection authentication, one-time passwords, and security questions, to
verify the legitimacy of credit card users.
5. Develop real-time monitoring capabilities: Develop real-time monitoring capabilities to

detect and flag any suspicious transactions as they occur.
6. Update and adapt to new types of fraud: Continuously update and adapt the system to
new types of fraud as they emerge, by monitoring the latest fraud trends and incorporating
them into the algorithms.
7. Test and refine the system: Test and refine the system by running it through a
series of simulated fraud scenarios and evaluating its effectiveness in detecting and
preventing fraudulent activity.
8. Deploy the system: Deploy the system in a real-world environment, such as a credit card
company or financial institution, to provide increased security to credit card users.
9. Monitor and maintain the system: Continuously monitor and maintain the system to
ensure it remains effective and up-to-date with the latest fraud prevention techniques.
5.2 TEST SETUP

1. Logical Regression

Logistic regression is a Machine Learning classification algorithm that is used

to predict the probability of certain classes based on some dependent variables. In short, the
logistic regression model computes a sum of the input features (in most cases, there is a bias
term), and calculates the logistic of the result. The output of logistic regression is always
between (0, and 1), which is suitable for a binary classification task. The higher the value, the
higher the probability that the current sample is classified as class=1, and vice versa.
2. Random Forest Classifier
Random Forest is a classification algorithm used in machine learning that builds

multiple decision trees by randomly selecting a subset of the features and samples in the
training dataset. Each decision tree is then used to make a prediction for a new data point, and
the overall output is determined by aggregating the results of all the individual decision trees
using majority voting. The final prediction is the class that receives the most votes from the
decision trees. By using multiple decision trees, Random Forest reduces overfitting and
increases the accuracy of the predictions. It also allows the algorithm to handle missing data
and maintain performance even when some of the features are irrelevant or noisy. Increasing
the number of trees in the forest can further improve the accuracy of the classification, but it
also increases the computational cost.
3. Decision Tree Algorithm
A decision tree is a machine learning technique that is used for both classification and
regression problems. It is a hierarchical structure that resembles a tree where each internal
node, including the root node, represents a "test" on a dataset's instances' attributes. The
results of each test are represented by the corresponding branches, which connect to further
internal nodes or leaf nodes, which stand for the class labels. To execute a decision tree, three
factors are required. First is the Target Attribute, which represents the class label. Second is
the Instances, which is a collection of instances for which the class label is already known.
Third is the Attributes List, which is a collection of predictor attributes used to create the
decision tree. The decision tree algorithm works by splitting the dataset into smaller subsets
based on the attribute that provides the best split. The attribute with the highest information
gain is selected as the root node for the tree, and the process is repeated recursively for each
child node until a stopping criterion is met.
Decision trees are commonly used in various applications, including spam detection, credit
risk analysis, and medical diagnosis. They are popular due to their simplicity, interpretability,

and

ability to handle both numerical and categorical data.
4. Naïve Bayes Algorithm
The approach presented here assumes that each feature contributes equally to the
classification objective and that all attributes are statistically independent, which may not
always be applicable in practice. To address this, the Bayes rule is employed to compute the
probability of an input belonging to each possible class and assigning the predicted class
based on the one with the highest probability. This allows the system to make more accurate
predictions even when the independence assumption between attributes does not hold. The
Bayes rule is a probabilistic model that considers prior knowledge of the probability of each
class and likelihood of the observed features to compute the probability of an input belonging
to each class.
5.3. TEST CASES
When both OTP verification and face recognition fail, the transaction is considered
unsuccessful. This design enforces the need for proper authentication in both factors,
highlighting the importance of validating the OTP and ensuring that the user's face matches the
registered identity. This approach enhances security and safeguards user's accounts from
potential fraud or unauthorized access.
If both OTP verification and face recognition are successful, indicating that the entered OTP
matches the generated one and the captured face matches the registered face, the transaction
proceeds. The system approves the transaction, allowing the necessary transaction to get
completed. This ensures that only authorized users who hold the correct OTP and match the
registered face can successfully complete transactions, minimizing the risk of fraudulent
activities and when OTP verification fails, regardless of the success of face recognition, the
transaction is considered unsuccessful. This design ensures that the correct OTP is provided as
an essential prerequisite for successful transaction authentication and reinforces the system's
security measures and when OTP verification is successful but face recognition fails, the
transaction is considered unsuccessful. This design emphasizes the importance of face
recognition as a critical authentication factor, even if the OTP verification is successful. By
requiring successful authentication in both factors we make the system more accurate and
secure.

Table 5.3 Table of Test Cases
OTP FACE RECOGNITION OUTPUT
(Transaction Successful-Yes/No)
PASS PASS YES
FAIL PASS NO
PASS FAIL NO
FAIL FAIL NO

CHAPTER 6
RESULTS AND DISCUSSIONS
The proposed project aims to develop a dependable system for detecting fraudulent credit card
transactions and authenticating credit card users. The system will leverage various machine
learning algorithms such as decision trees, random forests, and more, to scrutinize transactional
data and identify any suspicious activity. Additionally, the system will utilize diverse
authentication methods including facial detection authentication, one-time passwords, and
security questions to verify the legitimacy of the credit card user.
The machine learning algorithms used will be trained on a dataset consisting of past fraudulent
activities to recognize patterns and detect similar fraudulent transactions. By using machine
learning algorithms and statistical techniques to scrutinize transactional data, the system can
identify patterns and anomalies that are indicative of fraudulent activity. The authentication
methods used will help to verify the identity of the credit card user, reducing the likelihood of
unauthorized use. This will significantly increase credit card transaction security by detecting and
preventing fraudulent activities before they occur.

Fig. 6.1 Heatmap

This heatmap shown in our project is a visual representation of data that uses different colours to
depict values. It can be used to analyse various aspects of credit card usage and transactions. One
common application is to examine the difference in the variable’s patterns, where the heatmap can
represent different spending categories or time periods. This is mainly used in our project for helping
credit card companies identify unusual transaction patterns or outliers that may indicate fraudulent
activity. Furthermore, credit card issuers may employ heatmaps to assess risk by analysing historical
data, allowing them to identify high-risk areas or industries.
They provide a clear and intuitive way to understand patterns, trends, and anomalies in credit card
fraudulent transactions.

The system will also utilize diverse authentication meth6ds including facial detection
authentication, one-time passwords, and security questions to verify the legitimacy of the credit card
user. These methods will help to ensure that only legitimate credit card users are able to make
transactions, reducing the likelihood of unauthorized use.The implementation of this system is
expected to lead to substantial cost savings for both consumers and financial institutions. By reducing
the number of fraudulent transactions, financial institutions will be able to save money on
chargebacks, fraud investigations, and other related expenses. Additionally, consumers will be able to
avoid losses associated with fraudulent transactions.
The proposed system represents an effective solution to the problem of credit card fraud. By using
machine learning algorithms and diverse authentication methods, the system can significantly
increase credit card transaction security, prevent fraudulent activities by recognizing and
authenticating fraudulent activities before they happen, and lead to substantial cost savings for both
consumers and financial institutions.
6.1 Screenshots of the output

This page of the project acts as our homepage and allows us to fill the fields like Name ,City,
Zip code, State , Address ,CVV number and Expiration date that are necessary to make payments. It
gives us a placeholder with grey text being present as they can give an indication to the user about
what type of text must be present.

Fig 6.2 User Interface with indicators
This page interface gives us the field attributes that are taken and what needs to be
performed ranging from 1-28 variable attributes which are taken from a website named Kaggle and
trained only the Amount is the attribute that needs to be entered the other values are directly taken as
input.
Fig 6.3 User Interface with criteria to be filled
OTP (One-Time Password) verification is a crucial component in credit card fraud

detection programs. By integrating OTP verification into the system, an extra layer of security is
added to authenticate transactions and prevent unauthorized card usage. When a suspicious
transaction is

flagged, the program can generate and send an OTP to the cardholder's registered mobile number or
email address. The user then enters this unique OTP within a specified time limit to validate the
transaction. If the OTP is entered correctly, the transaction moves to the next step of authentication ;
otherwise, it is declined. This verification process ensures that even if a fraudster manages to obtain
credit card details, they would still require the OTP to complete the transaction, reducing the risk of
fraudulent activity. OTP verification significantly enhances the security of credit card transactions
and strengthens the overall fraud detection program by adding an additional authentication step to
mitigate the risk of unauthorized card usage.
Fig 6.4 OTP Authentication
When facial recognition fails as part of a two-way authentication system in credit card
transactions, there are measures that can be taken to ensure security. While doing this we can prevent
the transaction from happening before the fraudulent activity takes place. This helps to establish the
legitimacy of the transaction and prevents unauthorized access. By combining OTP verification with
facial recognition, the system adds an extra layer of security, and in cases where facial recognition
fails the transactions gets blocked.

Fig 6.5 Failure of face authentication
When facial recognition is successfully utilized as part of a two-way authentication system in

credit card transactions, it enhances security and provides several benefits. Once the user initiates a
transaction, their face is captured and compared to a reference image through facial recognition
algorithms. If the facial recognition process is successful, along with a valid OTP verification, the
transaction is approved. The benefits of this two-way authentication approach are two-fold. First, it
adds an additional layer of security by combining something the user knows (OTP) with something
they possess (their face). This significantly reduces the risk of unauthorized access or fraudulent
transactions. Second, it enhances user convenience by offering a streamlined and efficient
authentication process. Users do not need to remember complex passwords or carry additional
physical tokens, as their face and OTP serve as secure authentication factors. Overall, the integration
of OTP and facial recognition as a two-way authentication method in credit card transactions
improves security, mitigates fraud risks, and provides a convenient and user-friendly experience.
Fig 6.6 Successful transaction
When a user initiates a transaction, the system generates a unique OTP and sends it to the
registered mobile number's telegram as we do not currently have access to use the SMS feature on

python. The user is expected to receive the OTP and enter it into the system to validate their identity.
However, in this scenario, the OTP verification process fails, indicating that the entered OTP does
not match the one generated by the system or is incorrect which blocks the transaction from
happening.
Fig 6.7 Failure of OTP authentication
BotFather is a service provided by Telegram that allows users to create and manage their own
Telegram bots. It serves as an official bot that simplifies the process of bot creation by providing a
user-friendly interface. With BotFather, users can easily create their bots by specifying a bot name
and username. They can configure various settings such as profile picture, description, and default
commands. Additionally, BotFather enables users to manage their bots by allowing updates to the
bot's details and settings. By leveraging the Telegram Bot API, bots created using BotFather can
interact with users, perform automated tasks, and provide services within the Telegram platform.
Overall, BotFather streamlines the process of bot creation and management, empowering users to
harness the potential of Telegram bots for a wide range of applications.

Fig 6.8 OTP received by User
Telegram provides a secure and user-friendly platform for OTP (One-Time Password)
verification. When OTP verification is implemented in Telegram, the process typically involves the
following steps. First, the user initiates the OTP verification by providing their phone number or
relevant information. Next, the server generates a unique OTP for the user's verification. The OTP is
then securely delivered to the user's Telegram account as a message. Telegram's end-to-end
encryption ensures the confidentiality of the OTP during transmission. The user receives the OTP
message in their Telegram app, where they can easily access and enter the OTP into the verification
interface. Telegram's intuitive interface and real-time message delivery make the OTP verification
process seamless and efficient for users.
Additionally, Telegram's robust security measures provide reassurance that the OTP remains secure
and protected from unauthorized access. Overall, implementing OTP verification in Telegram offers
a reliable and convenient method to verify user identities while maintaining a high level of security.

Fig 6.9 BOTFather-Telegram Bot
Despite the successful OTP verification, the failure of face recognition raises concerns about the
authenticity of the user's identity. Face recognition serves as an additional layer of security to ensure
that the person initiating the transaction matches the registered face associated with the account.
When face recognition fails, it indicates that the user's face does not meet the expected criteria,
potentially indicating a fraudulent or unauthorized transaction attempt.As a result, the system
considers the transaction as unsuccessful and declines it. This decision prioritizes the accuracy and
integrity of the face recognition process as a crucial authentication factor. By rejecting the transaction
when face recognition fails, the system aims to prevent fraudulent transactions and unauthorized
access to the user's account. This approach ensures that successful OTP verification alone is not
sufficient to authorize a transaction. The combination of both OTP verification and face recognition
adds an extra layer of security, ensuring that the user's identity is validated through

multiple factors.

Fig 6.9 Face Recognition Failed
By training on a dataset consisting of past fraudulent activities, the system will recognize
patterns and detect similar fraudulent transactions. This approach will enable the system to detect
fraudulent activities in real-time, preventing further losses to the credit card companies and their
customers. The system will also utilize diverse authentication methods, including facial detection
authentication, one-time passwords, and security questions, to verify
the legitimacy of the credit card user. By using multiple authentication methods, the system can
prevent unauthorized access to credit card accounts and minimize the risk of fraudulent activities.

Fig 6.7 Face Recognition Unknown
In this picture, we go through with the OTP verification but when the face recognition is
supposed to happen we see that the face is not of the cardholder hence when the process of
recognition starts we see that the structure of the face doesn't match with the registered users face
hence its classified as an unknown structure.

CHAPTER 7
CONCLUSION
In conclusion, Credit card fraud is a prevalent problem in the financial sector, and it is
essential to develop a dependable system that can detect and prevent such activities. The proposed
system aims to address this issue by utilizing various machine learning algorithms and
authentication methods. The system's machine learning algorithms, including decision trees,
random forests, and more, will scrutinize transactional data and identify any suspicious activity.
By training on a dataset consisting of past fraudulent activities, the system will recognize patterns
and detect similar fraudulent transactions. This approach will enable the system to detect
fraudulent activities in real- time, preventing further losses to the credit card companies and their
customers. The system will also utilize diverse authentication methods, including facial detection
authentication, one-time passwords, and security questions, to verify the legitimacy of the credit
card user. By using multiple authentication methods, the system can prevent unauthorized access
to credit card accounts and minimize the risk of fraudulent activities. The proposed system's
implementation is expected to have a significant impact on credit card transaction security. By
detecting and authenticating fraudulent activities before they happen, the system can prevent
potential financial losses for both consumers and financial institutions. Moreover, the system can
help to reduce the cost of fraud investigations and reimbursements, leading to substantial cost
savings for financial institutions. The proposed system's development is crucial to address the
rising problem of credit card fraud. The system's utilization of various machine learning
algorithms and authentication methods can significantly increase credit card transaction security,
prevent fraudulent activities, and lead to substantial cost savings for both consumers and financial
institutions.
7.1 Scope for Future Enhancement
In order to further improve the proposed system's capabilities, advanced machine learning
algorithms and techniques such as deep learning and natural language processing could be
incorporated to better analyze and interpret transactional data. This would allow the system
to detect even more subtle patterns and anomalies that may indicate fraudulent activity. Moreover,
the system could be integrated with real-time fraud detection mechanisms that can monitor
transactions as they occur, identifying and flagging suspicious activities in real-time. This would
help to prevent fraudulent transactions from being processed before they can cause any harm.

Another potential enhancement to the system would be the incorporation of blockchain

technology.

By creating an immutable and transparent record of all transactions, the system could significantly
enhance security and prevent fraudulent activities. The use of blockchain technology would
provide an additional layer of security, making it much more difficult for fraudsters to tamper
with transactional data. Finally, the system could be extended to support more diverse payment
methods and authentication mechanisms. This would allow the system to accommodate the
evolving needs of consumers and businesses, while providing even greater protection against
credit card fraud. Overall, the proposed enhancements are expected to significantly improve the
effectiveness and efficiency of the system, leading to even greater savings for consumers and
financial institutions while further reducing the risk of credit card fraud. By constantly evolving
and adapting to new challenges, the system will be able to stay ahead of fraudsters and provide a
secure and reliable platform for credit card transactions.

REFERENCES
[1] Parth Roy, Prateek Rao, Jay Gajre, "Comprehensive Analysis for Fraud Detection of
Credit Card through Machine Learning" ,2021 International Conference on Emerging
Smart Computing and Informatics (ESCI) , March 2021.
[2] Ishika Sharma, Shivjyoti Dalai, Venktesh Tiwari, Ishwari Singh , Seema Kharb, "Credit
Card Fraud Detection Using Machine Learning & Data Science" , International Research
Journal of Engineering and Technology (IRJET) Vol. 09, Issue 06, Jun 2022.
[3] Anuruddha Thennakoon, Chee Bhagyani, Sasitha Premadasa, Shalitha Mihiranga, Nuwan
Kuruwitaarachchi ,"Real-time Credit Card Fraud Detection Using Machine
Learning",IEEE 9th International Conference on Cloud Computing, Data Science &
Engineering, 2019.
[4] H. Najadat, O. Altiti, A. A. Aqouleh, and M. Younes, ‘‘Credit card fraud detection based
on machine and deep learning,’’ in Proc. 11th Int. Conf. Inf. Commun. Syst. (ICICS),
Apr. 2020.
[5] S. Makki, Z. Assaghir, Y. Taher, R. Haque, M.-S. Hacid, and H. Zeineddine, ‘‘An
experimental study with imbalanced classification approaches for credit card fraud
detection,’’ IEEE Access.
[6] D. Tanouz, R Raja Subramanian, D. Eswar, G V Parameswara Reddy, A. Ranjith kumar,
CH V N M praneeth, "Credit Card Fraud Detection Using Machine Learning ", Fifth
International Conference on Intelligent Computing and Control Systems , 2021.
[7] MJ Madhury, H L Gururaj, B C Soundarya, K P Vidyashree ,B Rajendra , "Exploratory
analysis of credit card fraud detection using machine learning techniques",Elsevier B.V.
2022 .
[8] Fawaz Khaled Alarfaj , Iqra Malik , Hikmat Ullah Khan , Naif Almusallam , Muhammad
Ramzan , And Muzamil Ahmed ,"Credit Card Fraud Detection Using State-of-the-Art
Machine Learning and Deep Learning Algorithms", Deanship of Scientific Research at
Imam Mohammad Ibn Saud Islamic University through the
Research 2022.
[9] Manjeevan Seera, Chee Peng Lim, Ajay Kumar, Lalitha Dhamotharan, Kim Hua Tan ,"An
intelligent payment card fraud detection system", Springer Science+Business Media,
LLC, part of Springer Nature 2021.
[10] Sailusha Ruttala; Gnaneswar V.; Ramesh R.; Rao, G. Ramakoteswara, "Credit Card Fraud

Detection Using Machine Learning", International Conference on Intelligent Computing

Computing and Control Systems ,IEEE Xplore 2020.
[11] Shiyang Xuan,Guanjun Liu, Zhenchuan Li,Shuo Wang, Lutao Zheng, Changjun
Jiang,"Random forest for credit card fraud detection" ,IEEE, 2018.
[12] Subramanian R.R., Seshadri K. (2019) Design and Evaluation of a Hybrid Hierarchical
Feature Tree Based Authorship Inference Technique. In: Kolhe M., Trivedi M., T iwari
S., Singh V. (eds) Advances in Data and Information Sciences. Lecture Notes in
Networks and Syst ems, vol 39. Springer, Singapore
[13] Joshva Devadas T ., Raja Subramanian R. (2020) Paradigms for Intelligent IOT
Architecture. In: Peng SL., Pal S., Huang L. (eds) Principles of Internet of Things (IoT)
Ecosystem: Insight Paradigm. Intelligent Systems Reference Library, vol 174. Springer.
[14] R. R. Subramanian, B. R. Babu, K. Mamta and K. Manogna, "Design and Evaluation of a
Hybrid Feature Descriptor based Handwritten Character Inference T echnique," 2019
IEEE International Conference on Intelligent Techniques inControl, Optimization and
Signal Processing (INCOS), Tamilnadu, India, 2019, pp. 1-5.
[15] Andrew. Y. Ng, Michael. I. Jordan, "On discriminative vs. generative classifiers: A
comparison of logistic regression and naive bayes", Advances in neural information
processing systems, vol. 2, pp. 841-848,2002


Full Report Credict Card Fraud

Uploaded by

Copyright:

Available Formats

Full Report Credict Card Fraud

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Full Report Credict Card Fraud

Uploaded by

Copyright:

Available Formats

ABSTRACT

CHAPTER 2 – LITERATURE SURVEY………………………………………. 15

CHAPTER 3 – SYSTEM REQUIREMENT SPECIFICATION……………… 24

CHAPTER 4 – SYSTEM DESIGN ……...………………………………………. 29

CHAPTER 6 – RESULTS AND DISCUSSIONS…...………………………….. 38

Fig No Fig Name Page No.

Table No. Table Name Page. No.

Dept of CSE, SJBIT 2023-24 Page 6

Dept of CSE, SJBIT 2023-24 Page 7

Dept of CSE, SJBIT 2023-24 Page 9

1.2 Existing System

Dept of CSE, SJBIT 2023-24 Page 11

Dept of CSE, SJBIT 2023-24 Page 12

1.3 Proposed System

Dept of CSE, SJBIT 2023-24 Page 13

Dept of CSE, SJBIT 2023-24 Page 14

behaviors of individual customers.

1.5 Features with Scope

Dept of CSE, SJBIT 2023-24 Page 18

1.7 Organization of report

Dept of CSE, SJBIT 2023-24 Page 19

2. 1 General working features of the existing system

Dept of CSE, SJBIT 2023-24 Page 20

2.2 Different Types

Dept of CSE, SJBIT 2023-24 Page 21

Dept of CSE, SJBIT 2023-24 Page 22

Dept of CSE, SJBIT 2023-24 Page 24

Dept of CSE, SJBIT 2023-24 Page 26

Dept of CSE, SJBIT 2023-24 Page 28

2.3 Related papers in table format

Table 2.3 Literature Survey

Sl. Paper Author, Publisher Problem Technologi Outcome

Dept of CSE, SJBIT 2023-24 Page 29

4 Credit D. Tanouz, R Raja Insufficient Random It couldn't reach

7 An Manjeevan Seera, Finding out Support The best AUC

Dept of CSE, SJBIT 2023-24 Page 30

Springer Nature statistically recorded the

9 Credit Sailusha Ruttala; The results Random Random Forest

2.4 Technological Issues

Dept of CSE, SJBIT 2023-24 Page 31

Dept of CSE, SJBIT 2023-24 Page 32

SYSTEM REQUIREMENT SPECIFICATION

 Assumptions and Dependencies: This section identifies any assumptions or dependencies

Dept of CSE, SJBIT 2023-24 Page 33

3.1 Analysis / Feasibility Study

 Technical feasibility: This involves evaluating whether the proposed approach is

 Operational feasibility: This involves evaluating whether the proposed approach is

 Economic feasibility: This involves evaluating the cost-effectiveness of the proposed

Dept of CSE, SJBIT 2023-24 Page 34

3.2. Hardware Requirement Specification

Dept of CSE, SJBIT 2023-24 Page 36

Dept of CSE, SJBIT 2023-24 Page 37

Dept of CSE, SJBIT 2023-24 Page 38

4.1 Architectural Representation

Dept of CSE, SJBIT 2023-24 Page 39

Fig 4.1 Flow Chart of the proposed model

Dept of CSE, SJBIT 2023-24 Page 40

Dept of CSE, SJBIT 2023-24 Page 41

processed or cancelled. Thereby preventing the fraudulent transactions.

Fig 4.2 Block Diagram of the proejct