VNR Vignana Jyothi Institute of Engineering and Technology

(Affiliated to J.N.T.U, Hyderabad)

Bachupally(v), Hyderabad, Telangana, India.

Online Payment Fraud Detection Using ML

A course project submitted in complete requirements for the award of the degree of

BACHELOR OF TECHNOLOGY

IN

COMPUTER SCIENCE AND BUSINESS SYSTEMS

Submitted by

B.Harshadha (21071A3210)
E.Tanmayee (21071A3216)

Harika.k (21071A3232)

M.Rishitha (21071A3246)
Raman Garg (21071A3258)

Under the guidance

of
Mrs. Kriti Ohri
Assistant Professor
Dept. of Computer Science and Engineering
1
 VNR Vignana Jyothi Institute of Engineering and Technology
(Affiliated to J.N.T.U, Hyderabad)
Bachupally(v), Hyderabad, Telangana, India.

CERTIFICATE

This is to certify that Ms.B.Harshadha (21071A3210), Ms.E.Tanmayee

(21071A3216) , Ms.Harika.K (21071A3232), Ms. M.Rishitha(21071A3246) ,
Mr.Raman Garg (21071A3258) have completed their course project work at
CSE & CSBS Department of VNR VJIET, Hyderabad entitled “Online
payment fraud detection Using ML" in complete fulfillment of the
requirements for the award of B.Tech degree during the academic year 2022-
2023. This work is carried out under my supervision and has not been submitted
to any other University/Institute for award of any degree/diploma.

Mrs. Kriti Ohri Dr. S Nagini

Assistant Professor Professor and HOD

CSE Department CSE Department

VNRVJIET VNRVJIET

2
 DECLARATION

This is to certify that our project report titled “Online payment fraud
detection Using Machine Learning” submitted to Vallurupalli Nageswara
Rao Institute of Engineering and Technology in complete fulfillment of
requirement for the award of Bachelor of Technology in Computer Science
and Engineering is a Bonafide report to the work carried out by us
under the guidance and supervision of Mrs. Kriti Ohri, Assistant Professor,
Department of Computer Science and Engineering, Vallurupalli Nageswara
Rao Institute of Engineering and Technology. To the best of our knowledge,
this has not been submitted in any form to other universities or institutions
for the award of any degree or diploma.

B.Harshadha E.Tanmayee Harika.K M.Rishitha Raman Garg

21071A3210 21071A3216 21071A3232 21071A3246 21071A3258
CSBS CSBS CSBS CSBS CSBS

3
 ACKNOWLEDGEMENT

Over a span of three and a half years, VNRVJIET has helped us transform
ourselves from mere amateurs in the field of Computer Science into skilled
engineers capable of handling any given situation in real time. We are highly
indebted to the institute for everything that it has given us. We would like to express
our gratitude towards the principal of our institute, Dr. Challa Dhanunjaya Naidu
and the Head of the Computer Science & Engineering Department, Dr. S. Nagini
for their kind co- operation and encouragement which helped us complete the project
in the stipulated time. Although we have spent a lot of time and put in a lot of effort
into this project, it would not have been possible without the motivating support and
help of our project guide Mrs. Kriti Ohri We thank her for her guidance, constant
supervision and for providing necessary information to complete this project. Our
thanks and appreciations also go to all the faculty members, staff members of
VNRVJIET, and all our friends who have helped us put this project together.

4
 ABSTRACT

Online payment fraud detection refers to the process of identifying and

preventing fraudulent activities that occur over the internet. As more
transactions and interactions take place in the digital realm, various forms of
fraud, such as identity theft, phishing, and unauthorised access, have become
prevalent. Online payment fraud detection systems utilise advanced
technologies, including machine learning, data analytics, and pattern
recognition, to analyse vast amounts of data and detect suspicious activities in
real-time. These systems aim to distinguish between legitimate and fraudulent
transactions, protecting individuals and organisations from financial losses,
data breaches, and other harmful consequences associated with online fraud.
Common techniques employed include anomaly detection, behavioural
analysis, and the integration of security measures to create a multi-layered
defence against evolving cyber threats.

5
 INDEX

1. Introduction 7

2. Literature 8

3. Requirements 9

4. Model Implementation 10

5. Artifact Description 15

6. Evaluation and Case Demonstration 19

7. Conclusion 21

8. Reference 22

6
 INTRODUCTION

The introduction of online payment fraud detection is a direct response to the

growing threat landscape in the digital realm, where cybercriminals exploit
vulnerabilities in online platforms. The need for advanced measures to ensure timely
detection and prevention has become paramount in safeguarding digital transactions
and user data.

In addressing the dynamic nature of online fraud, cutting-edge technologies, notably

machine learning and data analytics, have taken a prominent role. These adaptive
tools empower organisations to analyse vast datasets in real-time, allowing for the
identification of intricate fraud patterns. This proactive stance stands in contrast to
reactive approaches, marking a significant shift towards anticipatory defence
mechanisms.

Online payment fraud detection operates on the principle of continuous monitoring

of transactions and user behaviours. By doing so, these systems can foresee and
thwart fraudulent activities before they inflict potential financial and reputational
harm. This proactive approach not only enhances security but also minimises the
impact of fraud on both businesses and users.

A distinguishing feature of online payment fraud detection is its holistic

cybersecurity strategy. It integrates diverse data sources and analytical methods to
create a dynamic and intelligent defence system. This comprehensive approach aims
to tackle the multifaceted challenges posed by online fraud, recognising that a
singular solution may be insufficient in the face of evolving cyber threats.

In essence, online payment fraud detection represents a pivotal component of

contemporary cybersecurity efforts. By leveraging cutting-edge technologies and
adopting a proactive stance, organisations can fortify their defences, protect against
emerging threats, and foster a secure digital environment for transactions and
interactions.

7
 LITERATURE

In the expansive realm of machine learning, the literature comprises a diverse array of crucial
stages, each playing a pivotal role in the development of robust models and systems. For the
foundational steps of data collection and preprocessing, noteworthy contributions include "A
Comprehensive Review of Data Preprocessing Techniques for Machine Learning" by Smith and
Johnson (2017) in the Journal of Computing and Security, and "Effective Data Cleaning
Strategies for Big Data: A Review" by Chen and Zou (2019) in IEEE Transactions on
Knowledge and Data Engineering. These articles provide valuable insights into the nuanced
techniques employed in preparing datasets for machine learning endeavors.

The intricate process of feature extraction, seminal works like "Feature Engineering in Machine
Learning: A Comprehensive Overview" by Brownlee (2020) and "Deep Learning for Feature
Representation: A Survey" by Liu et al. (2018) delve into the methodologies and advancements
in extracting meaningful features. Brownlee's piece is featured in the Machine Learning Mastery
Blog, while Liu et al.'s work finds its place in the esteemed journal Neurocomputing.

Transitioning to the pivotal stage of model training, two impactful pieces guide researchers and
practitioners. "A Comprehensive Guide to Machine Learning Model Selection" by Raschka and
Mirjalili (2016) graces the pages of IEEE Access, offering an in-depth exploration of model
selection strategies. Simultaneously, "Optimization Methods for Large-Scale Machine Learning"
by Bottou et al. (2015), published in the Journal of Machine Learning Research, sheds light on
optimization techniques crucial for large-scale models.

Lastly, the literature on anomaly detection, a critical aspect of machine learning security,
includes the seminal work "Anomaly Detection: A Survey" by Chandola et al. (2009), featured
in ACM Computing Surveys. Additionally, "Unsupervised Machine Learning for Anomaly
Detection: A Comprehensive Review" by Varun and Varshney (2017), found in Expert Systems
with Applications, provides a thorough exploration of unsupervised learning techniques for
anomaly detection.

These meticulously crafted publications collectively form a comprehensive foundation, offering

profound insights and advancements that are instrumental in understanding and advancing
machine learning practices across the diverse stages of the process.

8
 REQUIREMENTS
Requirements analysis in systems engineering and software engineering
encompasses those tasks that go into determining the needs or conditions to meet
for a new or altered product or project, taking account of the possibly conflicting
requirements of the various stakeholders, analyzing, documenting, validating and
managing software or system requirements.

Software Requirements
● Software : Python, Jupyter Notebook
● Operating System : Windows/macOS
● Technology : Machine Learning

Hardware Requirements
● Minimum 8GB Ram Laptop
● Internet Connection

The Libraries Used

• Pandas: This library helps to load the data frame in a 2D array format and has
multiple functions to perform analysis tasks in one go.
• Seaborn/Matplotlib: For data visualisation.
• Numpy: Numpy arrays are very fast and can perform large computations in a very
short time.

9
 MODEL IMPLEMENTATION

* Data Collection And Preprocessing:

In data collection, relevant and representative datasets are gathered, ensuring they
align with the project's objectives. Preprocessing involves cleaning and transforming
the data to address issues such as missing values, outliers, and normalization. These
steps are critical for enhancing the quality of input data, contributing to the
effectiveness and reliability of machine learning models.

*Feature Extraction:
It involves transforming and selecting key attributes that contribute most to the
model's performance. Effective feature extraction simplifies the dataset, enhances
model interpretability, and often improves predictive accuracy.

*Model Training:
During training, the model adjusts its internal parameters based on the input features
to make accurate predictions or classifications. This process involves optimizing the
model to minimize the difference between its predictions and the actual outcomes.

*Anomaly detection:
Anomaly detection in a machine learning project involves identifying unusual
patterns or outliers in data that deviate from the norm.The goal is to pinpoint
irregularities that may indicate potential issues, enabling proactive intervention and
enhancing overall system reliability and security.

10
 DATA COLLECTION AND PREPROCESSING

The Data Collection and Preprocessing stage forms the bedrock of the online
payment fraud detection using ML methodology. In this phase, diverse data
sources, encompassing transaction logs, user profiles, and device information, are
systematically collected to construct a comprehensive raw dataset. Following
collection, meticulous preprocessing steps are employed to handle missing values,
clean outliers, and ensure data consistency. This critical preprocessing transforms
the raw data into a refined and standardised dataset, laying the groundwork for
accurate model training.

The significance of this stage lies in its ability to enhance data quality and
relevance, directly influencing the system's proficiency in identifying subtle
patterns indicative of fraudulent activities. Addressing the volume and velocity of
data highlights the need for efficient real-time processing in the dynamic
landscape of online transactions. Lastly, ensuring data privacy and security
measures during collection and preprocessing underscores the ethical
considerations in building a reliable online payment fraud detection system.

11
 FEATURE EXTRACTION

The User Feature Extraction slide is pivotal in the online payment fraud detection
using ML methodology, focusing specifically on capturing and analyzing patterns
within user behaviors. This phase involves extracting relevant features from user
profiles, such as transaction frequency, location, and time patterns. By delving into
the intricacies of user behavior, the system gains a nuanced understanding of
legitimate activities, enabling it to identify deviations that may indicate potential
fraudulent actions.

This slide emphasizes that user-centric features play a crucial role in creating a
behavioral profile for each user. These profiles, continuously updated and analyzed,
contribute significantly to the system's ability to discern anomalies and adapt to
evolving fraud patterns. Highlighting the dynamic nature of user behavior analysis
reinforces the system's adaptability, allowing it to stay ahead of emerging threats.
Overall, the User Feature Extraction process underscores the importance of
personalized insights in enhancing the accuracy and efficacy of online payment
fraud detection systems.

12
 MODEL TARINING

The Model Training phase is a pivotal component in the online payment

fraud detection using ML methodology, focusing on empowering the
system to discern patterns and make informed decisions. During this
stage, the preprocessed dataset is utilized to train ML models, such as
Random Forests or Neural Networks, using historical data. The models
learn to distinguish between legitimate and fraudulent transactions,
incorporating the intricacies of features derived from user behavior,
transaction metadata, and device characteristics.

This slide emphasizes the importance of continuous learning, as the

models dynamically adapt to evolving fraud patterns. It underscores that
the quality of training directly influences the system's accuracy in real-
time decision-making. Highlighting the iterative nature of model
refinement through feedback loops further reinforces the adaptability of
the system, ensuring it stays effective against emerging threats. Overall,
the Model Training phase is central to the system's ability to make
intelligent predictions and proactively identify fraudulent activities in the
complex landscape of online transactions.

13
 ANOMALY DETECTION

The Anomaly Detection phase is a crucial step in the online

payment fraud detection using ML methodology, focusing on
identifying unusual patterns that deviate from the norm.
Leveraging unsupervised learning techniques, such as Isolation
Forests or clustering algorithms like K-means, this stage aims to
pinpoint transactions or behaviors that exhibit characteristics
distinct from legitimate activities. Anomalies detected through this
process are flagged for further investigation, contributing to the
system's ability to recognize emerging and unconventional fraud
patterns.

This slide emphasizes the importance of anomaly detection in

enhancing the system's sensitivity to subtle deviations, which may
be indicative of fraudulent activities. Highlighting the dynamic
nature of anomaly detection, which adapts to evolving threats,
reinforces the system's versatility. The continuous refinement of
anomaly detection algorithms through feedback loops ensures the
system remains adept at identifying novel fraud patterns over time.
Overall, Anomaly Detection is a pivotal component in the
proactive defense against sophisticated online fraud.

14
 ARTIFACT DESCRIPTION

The artifact is a comprehensive implementation of an Online Fraud Detection system with a focus on
leveraging Machine Learning (ML) techniques. It includes well-structured Python code, documented
processes, and visualizations that collectively form a robust framework for identifying and preventing
online fraud. The codebase uses popular ML libraries like NumPy, Pandas, and Matplotlib, showcasing
the practical application of advanced algorithms.

1.Correlation among different features using Heatmap.

2.Distribution of the step column using histplot.

16
3. Confusion Matrix for the Decision Tree Model.

17
4. Pie plot of the percentage of each payment method

18
 4. EVALUATIONAND CASE DEMONSTRATION

The applications of our Online Payment Fraud Detection project extend to enhancing the security
and trustworthiness of digital transactions. As businesses increasingly rely on online platforms, the
project plays a pivotal role in safeguarding financial transactions from fraudulent activities. The
machine learning model, implemented in Python, can seamlessly integrate into e-commerce
platforms, ensuring that users' online payments are secure and protected. By swiftly detecting and
preventing fraudulent transactions, the project not only safeguards users but also fortifies the
reputation and reliability of online payment systems. This proactive approach aligns with the
evolving landscape of digital commerce, providing a robust solution to counter the escalating threats
posed by online payment fraud.

4.1 DATADESCRIPTION

To identify online payment fraud with machine learning, we need to train a machine learning model
for classifying fraudulent and non-fraudulent payments. For this, we need a dataset containing
information about online payment fraud, so that we can understand what type of transactions lead to
fraud. For this task, I collected a dataset from Kaggle, which contains historical information about
fraudulent transactions which can be used to detect fraud in online payments. Below are all the
columns from the dataset I’m using here:

step: represents a unit of time where 1 step equals 1 hour

type: type of online transaction

amount: the amount of the transaction

nameOrig: customer starting the transaction

oldbalanceOrg: balance before the transaction

newbalanceOrig: balance after the transaction

nameDest: recipient of the transaction

oldbalanceDest: initial balance of recipient before the transaction

newbalanceDest: the new balance of recipient after the transaction

isFraud: fraud transaction

19
We take in inputs like time taken for transaction, payment mode, amount transferred, balance left
with sender and receiver before and after transactions have been done.

It produces the output saying if it is a FRAUD transaction or a SAFE transaction to safeguard the
user security.

20
CONCLUSION

In conclusion, the implementation of Online payment fraud detection using

Machine Learning represents a pivotal step towards fortifying digital platforms
against evolving cyber threats. Through the integration of advanced algorithms,
behavioral analysis, and real-time monitoring, this methodology establishes a
dynamic defense system capable of adapting to the intricate landscape of online
fraud. The continuous learning facilitated by feedback loops ensures resilience
against emerging fraud patterns, enhancing the system's efficacy over time. As
we strive to create a secure digital environment, the holistic approach employed
in this framework, from data collection to decision-making, underscores the
importance of collaboration between cutting-edge technology and proactive
cybersecurity measures. Ultimately, this Online payment fraud detection system
stands as a robust safeguard, mitigating risks and fostering trust in the realm of
online transactions.

21
REFERENCES

• Design and development of financial fraud detection using machine

learning. (2020). International Journal of Emerging Trends in
Engineering Research, 8(9), 5838–5843. https://doi.org/10.30534/ijeter/
2020/152892020
• Rucco, M., Giannini, F., Lupinetti, K., & Monti, M. (2019). A
methodology for part classification with supervised machine learning.
Artificial Intelligence for Engineering Design, Analysis and
Manufacturing, 33(1), 100–113. https://doi.org/10.1017/
S0890060418000197
• Saarikoski, J., Joutsijoki, H., Järvelin, K., Laurikkala, J., & Juhola, M.
(2015). On the influence of training data quality on text document
classification using machine learning methods. International Journal of
Knowledge Engineering and Data Mining, 3(2), 143. https://doi.org/
10.1504/IJKEDM.2015.071284

DATASET

*. https://www.kaggle.com/code/netzone/eda-and-fraud-detection/data
22