Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486

Machine Learning-Based Online Fraud

Detection and Prevention
Aravapalli Sri Chaitanya1 (Ph.D); Mohammad Tasneem Kowsar2; Dari
Udayasri3; Kanaparthi Ravikumar4; Borra Sai Ganesh5; Basa Datta Manikanta6
(ORCID : 0009-0008-5127-0250)1
1
(Assistant Professor) Artificial Intelligence & Machine Learning(CSE)
Dhanekula Institute of Engineering and Technology (JNTUK) Vijayawada, India
2
Artificial Intelligence & Machine Learning (CSE) Dhanekula Institute of Engineering and Technology
(JNTUK) Vijayawada, India
3
Artificial Intelligence & Machine Learning (CSE) Dhanekula Institute of Engineering and Technology
(JNTUK) Vijayawada, India
4
Artificial Intelligence & Machine Learning (CSE) Dhanekula Institute of Engineering and Technology
(JNTUK) Vijayawada, India
5
Artificial Intelligence & Machine Learning (CSE) Dhanekula Institute of Engineering and Technology
(JNTUK) Vijayawada, India
6
Artificial Intelligence & Machine Learning (CSE) Dhanekula Institute of Engineering and Technology
(JNTUK) Vijayawada, India

Publication Date: 2025/03/11

Abstract: As the rise of digital financial transactions exponentially increases, so does the need for efficient detection methods
for online fraud, a serious threat in the form of an online cybersecurity oker. In this paper, we proposed an AI-assisted
fraud detection framework involving supervised ML models (Logistic Regression, Random Forest, XGBoost) and
unsupervised anomaly detection (Isolation Forest, Autoencoders) for real-time fraud detection. Additional feature
engineering techniques, such as geolocation tracking, user profiling, and SMOTE for addressing class imbalance serve to
improve performance. For real-time monitoring, we use a Streamlit-based interface and deploy with Flask/FastAPI a
RESTful API of the trained model to easily integrate within the fintech. Precision, recall, F1-score and ROC-AUC metrics
for fraud risk assessment optimize fraud detection while placing a constraint on false positives, making a balanced fraud
risk assessment. The experimental results demonstrate their high accuracy and efficiency, which will guarantee this
framework a large scale for use in financial institutions, e-commerce and cyber security systems. The study introduces data-
driven solutions to beat evolving fraud; it helps enhance adaptive fraud detection for large systems.

Keywords: Fraud Detection, Machine Learning, Anomaly Detection, Online Transactions, Supervised Learning, Unsupervised
Learning, Financial Security, Cyber Threats.

How to Cite: Aravapalli Sri Chaitanya; Mohammad Tasneem Kowsar; Dari Udayasri; Kanaparthi Ravikumar; Borra Sai Ganesh;
Basa Datta Manikanta (2025). Machine Learning-Based Online Fraud Detection and Prevention. International
Journal of Innovative Science and Research Technology, 10(2), 1962-1968.
https://doi.org/10.5281/zenodo.14987486

I. INTRODUCTION account takeovers, phishing, and unauthorized transactions.

Conventional rule-based fraud detection systems generate
The rapid increase in digital financial transactions has led excessive false positives and fail to adapt to evolving fraud
to a surge in fraudulent activities, posing significant risks to patterns, highlighting the need for intelligent, data-driven fraud
financial institutions, e-commerce platforms, and online detection frameworks.
payment systems. With the widespread adoption of online
banking, digital wallets, and cashless transactions, Machine Learning (ML) and Anomaly Detection
cybercriminals exploit vulnerabilities in traditional fraud techniques have emerged as powerful tools to combat online
detection systems through techniques such as identity theft, fraud by leveraging historical transaction data to identify

IJISRT25FEB1491 www.ijisrt.com 1962

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486
suspicious patterns. Unlike traditional systems, ML algorithms This research aims to develop a scalable, AI-powered
continuously learn from data, improving fraud detection fraud detection system that continuously adapts to evolving
capabilities. Supervised learning algorithms such as Logistic fraud tactics, providing a robust security solution for financial
Regression, Random Forest, and XGBoost classify transactions institutions and digital payment services. With the ongoing
based on labeled fraud data, while unsupervised anomaly evolution of cyber threats, AI-driven fraud detection is
detection techniques like Isolation Forest and Autoencoders essential for safeguarding financial ecosystems and ensuring
detect novel fraud patterns by identifying deviations from the security of online transactions.
normal behavior. These approaches minimize false positives
and enhance fraud detection accuracy. II. LITERATURE SURVEY

Feature engineering plays a crucial role in fraud detection Machine learning methods have played a pivotal role in
by transforming raw transaction data into meaningful features fraud detection in multiple fields, with recent research
such as transactional behavior, geospatial tracking, user examining diverse models for increased accuracy and
profiling, and device fingerprinting.However, the class efficiency. U. Siddaiah [1] points to the efficiency of XGBoost
imbalance in fraud detection datasets poses a significant in fraud detection, with particular emphasis on the importance
challenge, as fraudulent transactions constitute only a small of Synthetic Minority Over-sampling Technique (SMOTE) in
fraction of overall data.Techniques such as Synthetic Minority class imbalance. He recommends further research on the
Oversampling Technique (SMOTE) address this imbalance by application of autoencoders in anomaly detection, which
generating synthetic samples of the minority class, improving would enhance fraud detection by learning representations of
model generalization and reducing bias. data in an unsupervised way. M. Naga Raju [2] also
emphasizes the effectiveness of Long Short-Term Memory
For real-time fraud detection, a web-based monitoring (LSTM) networks to extract sequential patterns, specifically in
system using Streamlit provides an intuitive interface for Unified Payments Interface (UPI) fraud cases. He
visualizing fraudulent activities and monitoring transaction recommends the creation of hybrid real-time models to
risks dynamically.The trained ML models are deployed as enhance fraud detection systems by combining various
RESTful APIs via Flask or FastAPI, enabling seamless approaches.
integration with financial institutions and e-commerce
platforms.This deployment ensures scalability and efficient Pinku Ranjan [3] assesses the Random Forest
fraud detection in high-volume transactional environments. performance, with a significant accuracy of 96.64%. His work
Real-time alert mechanisms notify financial institutions and emphasizes the need for data preprocessing to maintain
customers of suspicious transactions, enhancing security and balanced datasets, which is essential for reducing bias and
enabling prompt intervention. enhancing detection rates. Deepanshu Thapa [4] contends that
the XGB Classifier provides the highest precision at 99.95%,
Model performance is evaluated using precision, recall, emphasizing the critical role of feature engineering and
F1-score, and ROC-AUC to balance fraud detection sensitivity ensemble learning. He highlights the use of strong features and
and false positive reduction. Precision measures the proportion good optimized machine learning pipelines for real-time fraud
of correctly identified fraudulent transactions, recall assesses detection. Reem A. Alzahrani [5] emphasizes click fraud
the model’s ability to detect fraud cases, the F1-score balances detection using both machine learning and deep learning
precision and recall, and ROC-AUC evaluates the model's methodologies. Her paper mentions a 97.34% accuracy via
overall discriminatory power. Recurrent Neural Networks (RNN), exhibiting deep learning
capability against fraud prevention. She also points to the need
Fraud takes various forms, including financial fraud for adaptive learning and real-time analysis in order to offset
(credit card fraud, mortgage fraud), identity theft, payment the fast-developing fraudulent actions.
fraud (account takeovers, chargeback fraud), cyber fraud
(phishing, malware attacks), and corporate fraud (insider When it comes to fraud detection in e-commerce, Abed
trading, asset misappropriation).To counter fraud, Mutemi [6] discusses deep models, specifically for their
organizations employ rule-based systems, ML models, support of LSTM because it is efficient with sequential data.
behavioural analytics, anomaly detection, real-time His work recognizes the limitations of small real-world
monitoring, multi-factor authentication (MFA), and graph datasets and changing nature of fraud schemes, necessitating
analytics. enhanced data gathering and responsive measures. Rayene
Bounab [7]'s work in the area of detecting healthcare fraud
Despite advancements, fraud detection faces challenges uses SMOTE-ENN, an oversampling and noise elimination
such as high false positives and false negatives, evolving fraud strategy, to address class imbalance. Employing Decision
tactics, data privacy concerns, scalability issues, and Trees, her model attains an accuracy level of 99%, which
integration with legacy systems. Future fraud detection will proves the efficiency of tackling data imbalance problems. She
rely on AI-driven models, blockchain technology for also recommends future research to investigate deep learning
transparency, advanced biometric authentication, and federated methods, which can potentially enhance detection rates in the
learning for privacy-preserving fraud detection. Explainable AI healthcare industry.
(XAI) will also enhance interpretability and trust in fraud
detection decisions. Dr. Yogesh W. Bhowte [8] focuses on detecting financial
fraud, and he talks about the importance of feature engineering

IJISRT25FEB1491 www.ijisrt.com 1963

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486
and big data analytics in detecting fraudulent transactions. His the data, and classifies the transaction as fraudulent or genuine
work highlights the importance of scalable fraud detection based on the trained Random Forest model.The API is
systems that can handle enormous amounts of financial data in equipped with robust error-handling mechanisms (Reem
real time. Dr. Shaik Rehana Banu [9] proposes a hybrid A.Alzahrani [5]) to validate incoming data, ensuring smooth
Convolutional Neural Network (CNN)-RNN financial fraud operation even when faced with missing or malformed
detection model with an accuracy of 99.2%. She emphasizes inputs.Additionally, the system is designed with a high-
data preprocessing as being important in sharpening model throughput, low-latency architecture (Abed Mutemi [6]),
performance and eliminating false positives. Finally, making it highly scalable for financial environments where
Ibomoiye Domor Mienye [10] explores deep learning models transactions occur at an extremely high frequency.
including LSTM and Gated Recurrent Units (GRU) to detect
credit card fraud. His results show hybrid models improve To evaluate the performance of the model, various
detection performance, and future work lies in developing metrics of evaluation are utilized, such as confusion matrix,
methodologies to effectively mitigate class imbalance. These classification reports, and AUC-ROC curves (U.Siddaiah
studies overall highlight the paradigm-shifting effect of [1]).These metrics allow analyzing precision, recall, and F1-
machine learning on fraud detection, with new developments score to ensure that the model has a robust balance between
in deep learning, feature engineering, and real-time analysis fraud detection and avoiding false positives.Feature
leading the way towards more robust fraud prevention importance analysis is conducted to determine how various
solutions. variables play a role in fraud detection (M.Naga Raju [2]).In
addition, hyperparameter optimization through grid search and
III. PROPOSED SOLUTION random search maximizes model performance by identifying
the optimal parameters with the optimal balance of
The proposed fraud detection system is a strong machine computational efficiency and prediction power (Ibomoiye
learning-based system that is capable of detecting fraudulent Domor Mienye [10]).
financial transactions with great accuracy and efficiency.As
fraudulent activities are becoming more sophisticated, the IV. METHODOLOGY
system incorporates sophisticated ensemble learning
algorithms and efficient data preprocessing schemes to handle The fraud detection system adopts a systematic four-
huge amounts of financial data with ease.One of the most phase process: data preprocessing, model building,
important elements of the system is the Random Forest deployment, and testing, to have a thorough and technically
Classifier (Pinku Ranjan [3]), a type of ensemble model sound process. During the data preprocessing phase, raw
composed of many decision trees used to increase predictive financial transaction data is converted into structured form
power and reduce overfitting. The model is well-generalizing optimized for machine learning. This involves dealing with
across all the types of financial transactions and provides missing values, encoding categorical variables with Label
consistent fraud detection. Encoding, and feature selection. Categorical features like
transaction type and account numbers are numerically
The major difficulty in fraud detection is class converted to make them compatible with the Random Forest
imbalance—fraudulent transactions tend to be considerably model. SMOTE creates synthetic minority class instances to
less than authentic ones.To address this, the Synthetic overcome class imbalance, enhancing the model's capability to
Minority Over-sampling Technique (SMOTE) is utilized identify fraudulent patterns.
(Rayene Bounab [7]).SMOTE provides balancing to the
dataset by creating synthetic examples of fraudulent The model development stage emphasizes machine
transactions so that the model can better learn patterns of learning algorithm selection and optimization. Random Forest
fraud.The dataset is preprocessed carefully, and categorical Classifier is used based on its ensemble learning nature to avoid
features such as transaction type and account numbers are overfitting and increase predictive stability. The training
encoded as numerical values using Label Encoding procedure includes domain-guided feature selection and
(Deepanshu Thapa [4]).This conversion renders the dataset hyperparameter optimization with grid search and k-fold cross-
suitable for machine learning models.In addition, feature validation to fine-tune the number of estimators, maximum
engineering targets high-impact financial features with depth of trees, and splitting values.This optimization ensures
irrelevant attributes and IDs removed that might bring in bias high precision, recall, and F1-score for high fraud detection.
(Dr. Yogesh W.Bhowte [8]).The strategy guarantees the
model learns about transaction behaviors and not memorizing To ensure real-time deployment, the system is wrapped
particular instances. around a Flask-based RESTful API and seamlessly interacts
with financial systems. The API includes a /predict endpoint
In real-time fraud detection, the trained model is accepting JSON payloads, encoding categorical features,
implemented as a Flask-based RESTful API (Dr. Shaik normalizing numerical features, and passing the data to the
Rehana Banu [9]).This API harmoniously integrates into trained model. The API outputs a probabilistic class label for
financial systems and enables transactions to be assessed in transaction legitimacy, and it comes equipped with extensive
real-time.The /predict endpoint accepts formatted JSON error-handling functions to confirm received data.To ensure
payloads of transaction information.When a request is made, scalability, the system effectively handles high transaction
the system dynamically encodes categorical features, prepares volumes without compromising on latency.

IJISRT25FEB1491 www.ijisrt.com 1964

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486
Table 1Model Metrics
Algorithm Accuracy Precision Recall F1 Score
Logistic Regression 91.5 88.7 83.4 85.9
Decision Tree 93.2 90.1 86.8 88.4
Random Forest 97.6 96.2 94.1 95.1
XGBoost 98.3 97.5 95.8 96.6
SVM 95.1 94.3 89.7 91.9
K-Means 78.4 78.5 81.2 79.8

Model performance is assessed with a confusion matrix, L1 (Lasso) and L2 (Ridge) regularization to avoid overfitting
precision-recall, and SHAP values to facilitate interpretability. and enhance generalization. It also uses tree-pruning methods
Stress testing and load balancing ensure performance under like depth-wise and loss-guided growth, which minimize
stress, and retraining on the fly adjusts the model to changing unnecessary splits and improve interpretability. The algorithm
fraud patterns.This adaptive, scalable, and highly efficient also has missing value handling, in-built cross-validation, and
platform offers financial institutions a robust fraud prevention parallel processing, making it extremely scalable and ideal for
tool, minimizing risks and guaranteeing transaction integrity in complex machine learning applications where speed and
online finance. accuracy are needed.

V. ALGORITHMS  Random Forest

Random Forest is a strong ensemble learning method that
 Logistic Regression builds many decision trees and makes predictions by
Logistic Regression is a supervised algorithm that is aggregating their predictions to improve accuracy and
highly applied in binary classification, thereby being suitable stability. It employs bagging (Bootstrap Aggregating), training
for fraud detection. It labels financial transactions as fraud or each tree on a randomly sampled subset of the data with
genuine by considering important attributes such as the amount replacement. It also chooses a random subset of features at
of a transaction, geolocation, time stamp, and user activity. The every split, which decreases correlation between trees and
model applies the logistic (sigmoid) function to transform input enhances generalization. Predictions are made by majority
variables into a probability between 0 and 1, supporting voting for classification or averaging for regression.
threshold-based prediction. Transactions beyond a specified
value are labeled fraudulent, while the rest are tagged as y^=mode{T1(x),T2(x),...,Tn(x)}
genuine.
One of the primary benefits of Random Forest is that it
P(Y=1∣X)=1/1+e−(β0+β1X1+β2X2+...+βnXn) can manage high-dimensional data, noisy features, and
missing values and reduce overfitting. It also assigns feature
One of the main benefits of Logistic Regression is that it importance scores, facilitating interpretability and feature
is interpretable, with its coefficients reflecting the impact of selection. The method is insensitive to outliers and performs
each feature on predictions. This adds transparency to fraud well with structured and unstructured data. It can be
detection, which helps financial institutions in risk assessment. computationally costly, particularly with big datasets or large
Logistic Regression is also computationally efficient, which numbers of trees. Optimized hyperparameter tuning (e.g.,
makes it suitable for structured data and real-time fraud n_estimators, max_depth) is important. In spite of its
detection, where decisions need to be made quickly. Although complexity, Random Forest has a very good predictive
more sophisticated models such as decision trees and deep performance and is commonly applied to fraud detection,
learning are available, Logistic Regression is a good baseline medical diagnosis, and recommender systems.
because of its simplicity and performance.
 Decision Tree
 XGBOOST Decision Trees are interpretable and general-purpose
XGBoost or Extreme Gradient Boosting is an ensemble supervised learning algorithms for classification and
learning algorithm with high performance using gradient regression. They partition the feature space recursively into
boosting, targeted at regression and classification problems. homogeneous subsets based on decision rules derived from
XGBoost refines basic gradient boosting with new features criteria such as Gini impurity, entropy, or mean squared error.
such as regularization, parallelism, and improved tree Splitting is continued until a stopping criterion is reached, e.g.,
construction to make it efficient and fast. XGBoost builds maximum tree depth or minimum samples per leaf node.
decision trees sequentially, with each tree trying to fit the
mistakes of the last trees through optimization of a loss Entropy: −∑pi log2 pi
function using second-order Taylor approximation for
accurate updates. A significant strength of Decision Trees is their
interpretability, as they can easily visualize the decision-
Fm(x)=Fm−1(x)+γhm(x) making process. They can easily work with both numerical
and categorical data without the need for heavy preprocessing.
One of the biggest strengths of XGBoost is that it can They are, however, susceptible to overfitting, particularly if
efficiently process large-scale, high-dimensional data. It uses grown excessively deep, picking up noise from the data. The

IJISRT25FEB1491 www.ijisrt.com 1965

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486
danger can be avoided by pruning, applying constraints on decides whether the transaction is genuine or fraudulent. The
splits and leaf nodes, or ensembling with the help of analysis process is quick, using machine learning algorithms
techniques such as Random Forests and Gradient Boosted to identify possible fraud effectively. After the analysis, the
Trees. In spite of these difficulties, Decision Trees are still a system quickly returns a result, indicating to the user whether
valuable tool for representing intricate decision boundaries in the transaction is safe to continue or has been marked as
fraud detection, medical diagnosis, and recommendation fraudulent. This feedback in real-time prevents unauthorized
systems, among other applications. transactions and adds security. The smooth integration of the
fraud model guarantees fast and precise decision-making,
VI. RESULTS enabling enterprises and consumers to reduce risks and have
faith in financial transactions.
In the fielded environment, the fraud detection system is
designed to provide effective real-time analysis and timely
decision-making. The front-end of the system is a simple web-
based interface named "Check Your Data Secure," which
gathers critical transaction parameters via a formatted input
form. Once submitted, the information is passed to a back-end
API where a strong machine learning model analyzes the data.
This model relies on real-time feature transformation for
fraudulent/legitimate transaction classification purposes,
returning the decision as quick feedback to the user.
Deployment strategy promotes clean segregation of
presentation layer from processing engine for optimum
scalability as well as reliability during heavy load times. This
combined approach not only boosts the overall effectiveness
of the system but also optimizes data collection and fraud
analysis, enhancing the security of financial transactions
across the entire deployment structure.

Fig 2 Checking Fraud

After the user enters the transaction details and clicks the
Submit button, the fraud detection system analyzes the input
using advanced machine learning algorithms. It evaluates key
factors such as transaction amount, location, and user history
to determine if the transaction is legitimate or fraudulent. If
fraud is detected, the system immediately displays a warning
message, alerting the user that the transaction is suspicious.
This real-time detection helps prevent unauthorized activities,
ensuring security and trust in financial transactions. By
identifying fraudulent transactions instantly, the system
minimizes risks and enhances protection for both businesses
and users.

Fig 1 Deployment View

Once the user inputs the transaction information in the

corresponding fields and submits it, the fraud detection system
processes the data in real time. The system scans the input
through an advanced fraud detection model that is capable of
detecting suspicious patterns and anomalies. Through the
analysis of different factors like transaction value, location,
user history, and other important parameters, the model
Fig 3 Fraud Detection

IJISRT25FEB1491 www.ijisrt.com 1966

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486
On receipt of transaction information, the fraud detection (PCA) were used to reduce dimensionality while retaining
system undertakes real-time processing and anomaly informative features. Hyperparameter tuning with Grid Search
identification. Through sophisticated machine learning and Random Search also enhanced model performance.
algorithms, the system examines crucial transactional
parameters like amount, geolocation, device fingerprinting, For online fraud detection, the trained models were
user behavioral patterns, and historical transactions. The deployed through a Flask-based RESTful API and seamlessly
model utilizes statistical and AI-based fraud detection methods interfaced with financial transaction systems. There is a
like supervised learning and anomaly detection to identify the specialized /predict endpoint that receives transaction data,
legitimacy of the transaction. If no fraud is detected, the performs feature transformations, and provides fraud
system provides a validation response, validating the predictions, which are archived in a relational database
authenticity of the transaction and authorizing further (MySQL/PostgreSQL) for post-analysis. The infrastructure is
processing. This real-time feedback mechanism increases deployed on cloud resources (AWS/GCP) for scalability, high
security, providing smooth and risk-free financial transactions. reliability, and low-latency processing of transactions at high
The fraud detection system combines real-time monitoring, volume. Docker containerization was utilized for uniform
feature engineering, and predictive analytics to constantly deployment across environments.
enhance detection accuracy. Through AI-based decision-
making, the system successfully reduces false positives while Security features like data encryption, role-based access
keeping strong fraud prevention in place. This methodology control (RBAC), and API authentication protect transaction
allows for commercial and financial institutions to handle data. The project offers a scalable, high-performance fraud
legitimate transactions effectively, with high security, detection system integrating advanced machine learning, real-
operational dependability, and customer confidence. time API deployment, cloud infrastructure, and security best
practices, providing financial institutions with a
comprehensive solution for fraud prevention and digital
banking security.

REFERENCES

[1]. U. Siddaiah, P. Anjaneyulu, M. Ramesh, Y. Haritha.

"Fraud Detection in Online Payments using Machine
Learning Techniques"
.DOI:10.1109/ICICCS56967.2023.10142404.
[2]. M. Naga Raju, Yarramreddy Chandrasena Reddy,
Polavarapu Nagendra Babu, Venkata Sai Pavan
Ravipati, Velpula Chaitanya. "Detection of Fraudulent
Activities in Unified Payments Interface using
Machine Learning -LSTM Networks". DOI :
10.1109/ICCPCT61902.2024.10672890.
Fig 4 Not Fraud View [3]. Pinku Ranjan, Kammari Santhosh, Somesh Kumar,
Arun Kumar. "Fraud Detection on Bank Payments
VII. CONCLUSION Using Machine Learning". DOI :
10.1109/ICONAT53423.2022.9726104.
The fraud detection mechanism in this project is a [4]. Deepanshu Thapa, Aditya Harbola ,Aditya Joshi,
machine learning-driven solution that effectively identifies Vandana Rawat, Neha Pandey. "Machine learning
fraudulent financial transactions with minimal false positives. Models for Detecting Anomalies in Online Payment: A
It uses supervised and unsupervised learning algorithms such Comparative Analysis". DOI :
as Logistic Regression, Decision Trees, Random Forest, 10.1109/NMITCON58196.2023.10276124.
XGBoost, Support Vector Machines (SVM), and K-Means [5]. Reem A. Alzahrani, Malak Aljabri, Rami Mustafa, A.
Clustering. Supervised algorithms such as XGBoost and Mohammad. "Ad Click Fraud Detection Using
Random Forest make decisions based on labeled historical Machine Learning and Deep Learning Algorithms".
data, whereas unsupervised algorithms such as K-Means DOI : 10.1109/ACCESS.2025.3532200.
Clustering identify anomalies by recognizing patterns that [6]. Abed Mutemi, Fernando Bacao."E-Commerce Fraud
deviate from the normal pattern of transactions. Detection Based on Machine Learning Techniques:
Systematic Literature Review". DOI:
The data, derived from actual financial transactions, was 10.26599/BDMA.2023.9020023.
subjected to deep preprocessing, such as dealing with missing [7]. Rayene Bounab, Karim Zarour, Bouchra Guelib,
values, detection of outliers, normalization of numerical Nawres Khlifa."Enhancing Medicare Fraud Detection
features, and categorical feature encoding via Label Encoding Through Machine Learning: Addressing Class
and One-Hot Encoding. Feature engineering was utilized to Imbalance With SMOTE-ENN". DOI -
improve model performance. Class imbalance was dealt with 10.1109/ACCESS.2024.3385781.
using the Synthetic Minority Over-sampling Technique [8]. Dr. Yogesh W Bhowte, Dr. Arundhati Roy, Dr. K.
(SMOTE). Feature selection techniques like Recursive Bhavana Raj, Dr. Megha Sharma, Dr. K. Devi, Dr.
Feature Elimination (RFE) and Principal Component Analysis

IJISRT25FEB1491 www.ijisrt.com 1967

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14987486
Prem Latha Soundarraj."Advanced Fraud Detection
Using Machine Learning Techniques in Accounting
and Finance Sector". DOI -
10.1109/ICONSTEM60960.2024.10568756.
[9]. Dr. Shaik Rehana Banu, Dr. Taviti Naidu Gongada,
Kathari Santosh, Harish Chowdhary, Sabareesh R, Dr.
S. Muthuperumal."Financial Fraud Detection Using
Hybrid Convolutional and Recurrent Neural Networks:
An Analysis of Unstructured Data in Banking". DOI -
10.1109/ICCSP60870.2024.10543545.
[10]. Ibomoiye Domor Mienye, Nobert Jere. "Deep Learning
for Credit Card Fraud Detection: A Review of
Algorithms, Challenges, and Solutions". DOI -
10.1109/ACCESS.2024.3426955.
[11]. Saad Hammood Mohammed, Abdulmajeed Al-
Jumaily, Mandeep S. Jit Singh, Víctor P. Gil Jiménez,
Aqeel S. Jaber, Yaseein Soubhi Hussein, Mudhar
Mustafa Abdul Kader Al-Najjar, Dhiya Al-Jumeily."A
Review on the Evaluation of Feature Selection Using
Machine Learning for Cyber-Attack Detection in Smart
Grid". DOI - 10.1109/ACCESS.2024.3370911.

IJISRT25FEB1491 www.ijisrt.com 1968