Aims and Objectives
Aims:
The study aims to develop a “credit card fraud detection system”by utilising “machine learning algorithms” using Python programming language.
Objectives:
- To investigate existing “credit card fraud detection” techniques and methodologies for identifying strengths and limitations of the existing systems
- To gather and pre-process a thorough dataset for developing “credit card fraud detection”
- To design and implement appropriate “machine learning models” for developing credit card fraud detection systems using Python
- To formulate an effective credit card “fraud detection system” for addressing fraud patterns in credit card transactions
Proposed Plan of Work
Background and Rationale
Figure 1: Credit Card Fraud in the UK
(Source: Andrews, 2023)
According to Dastidar et al. (2022), credit card fraud detection can be treated as process of indentation of and rejection of fraudulent activities in credit card transactions. In 2022, there were almost 475,038 plastic card frauds and online transactions offences occurred in the UK, which has increased by more than 67% (Andrews, 2023). In the US, almost 389,827 credit card frauds were registered in the year 2021 (Andrews, 2023). According to the report by Gibbs (2022), cyberattackers target websites that handles enormous amount of “low-value transactions”. As per the view of Ileberi et al. (2022), primary reason for the ineffectiveness of existing credit card fraud detection systems is the class imbalance in credit card transactions. It has been observed that almost 99.8% of “credit card transactions” are normal transactions and only 0.02% of credit card transactions are fraudulent. This makes it difficult for detecting credit card transactions. Therefore, through the development of a “machine learning-based credit card detection system”, it will be possible to detect fraudulent activities in “credit card transactions”. This has acted as a source of motivation behind the incorporation of the research topic.
Preliminary Literature Review
The advancement of “e-payment system” have scintillated an enhancement in monetary fraud cases in “credit card” transactions. Hence, it has become emerging important to deploy machine learning-based mechanisms for detecting fraud activities. The research by Ileberi et al. (2022) has proposed a “machine learning (ML) based credit card fraud detection engine” through the use of the “genetic algorithm” for selecting features. Ileberi et al. (2022) have used multiple ML classifiers like Decision Tree, Logistic Regression, ANN, Naive Bayes, and many more for developing the detection engine. Trivedi et al. (2020) have incorporated Tree Classifier, neural networks, SVM, and XGBoost classifier strategies for the detection of unusual transactions on “skewed credit card fraud data sets”. As per the view of Maniraj et al. (2019), fraud detection includes monitoring activities of users to estimate and evaluate objectionable behaviour like intrusion, defaulting, and many more. Due to this, Maniraj et al. (2019) have deployed different “anomaly detection algorithms” like “Isolation Forest algorithms” and “Local Outlier factors” on “PCA transformed credit card transaction” data.
Sailusha et al. (2020) have implemented Adaboost algorithm, and random forest algorithm for detecting unusual activities. The results obtained from these two models are compared in terms of accuracy, recall value, F1 score, and precision. Sailusha et al. (2020) have found that the AdaBoost algorithm poses a greater accuracy and precision in detecting fraud. As per the view of Kim et al. (2019), screening of “fraudulent transactions” required authorisation of card issuers. Kim et al. (2019) have checked “high-risk transactions” from “data-driven scoring models” through the deployment of “deep learning method” and “hybrid ensemble” techniques. On the other hand, Thennakoon et al. (2019) utilised predictive analytics through the implementation of “machine learning models” along with the integration of API modules for deciding whether a credit card transaction is fraudulent or genuine. Within the research of Thennakoon et al. (2019), a novel strategy was introduced for addressing the “skewed distribution” of data.
The findings of Nelson report have stated that card losses across the world reached $35 billion in 2020 (Najadat et al. 2020). Therefore, it becomes severely important for service providers to ensure the identification of abnormal activities. Najadat et al. (2020) have implemented “BiLSTM- MaxPooling-BiGRU-MaxPooling” model on “IEEE-CIS Fraud Detection dataset” for detecting unusual activities in historical credit card transactions. Najadat et al. (2020) have achieved a model accuracy of approximately 91.37%.
Research Methodology
Research Approach
The research will be fundamentally based on secondary research. An “Inductive approach” will be taken into measure for evaluating secondary data collected from reliable sources. “Inductive approach” refers to the research process for developing patterns of theories and explanations from the observations of the researchers for achieving a general conclusion corresponding to the research problem. This approach is effective to develop a clear linkage between research objectives and the evaluation of the researchers from searching secondary theoretical data and information. This will help in analysing issues in existing “credit card fraud detection” systems.
Research Design
A “Quasi-experimental” design will be contemplated for evaluating existing research based on “credit card fraud detection” systems. “Quasi-experimental” design refers to an interventional study for developing a “cause and effect relationship” between the experimental data achieved by the researcher’s work. This provides accurate evidence by increasing transparency of the experimental process in the research without the considerations of randomization during data collection. Through the deployment of the “Quasi-experimental” design, it will be possible to develop an effective “credit-card fraud detection system” using “machine learning algorithms”.
Data collection Techniques
The secondary qualitative data will be gathered by applying secondary data collection technique. Reliable databases like “Google Scholar”, “ProQuest”, and many more will be utilised for collecting scholarly articles, journals, and other resources.
Software Specification and Tools
Python Programming language will be applied to developing different machine learning models. Through the deployment of different machine learning algorithms like “Decision Trees”, “Logistic Regression”, “Support Vector Machine”, “ANN”, and many more, a final “fraud detection system” will be developed.
Ethical Consideration
Maintaining data integrity and prevention of authorised access to collected data is going to be the primary challenge in this research. Therefore, in order to main data integrity and prevent unauthorised data accessibility, “UK Data Protection Act (2018)” will be taken into consideration.
Expected Outcomes
Through the execution of this research, it will be possible to develop a “machine learning-based fraud detection system”, which will help in the detection of outliers and abnormalities in credit card transactions.
Resources
- Computing Resources
- Knowledge of Python programming
- Machine Learning algorithms
- Jupyter Notebook software
The location of this project will be at University of Bolton.
Period of Study
Task Name | Duration | Start | Finish |
Initial Stage | 10 Days | 06.07.23 | 19.07.23 |
Identification of Topic | 3 Days | 20.07.23 | 23.07.23 |
Literature Review | 15 Days | 24.07.23 | 09.08.23 |
Methodology | 10 Days | 10.08.23 | 20.08.23 |
Result and Discussion | 7 Days | 21.08.23 | 28.08.23 |
Conclusion | 5 Days | 29.08.23 | 03.09.23 |
Figure 2: Detailed schedule of the program
The detailed outline of the program is demonstrated in the figure. The total time required for the complete execution of the project will be approximately 55 days. The overall research work will be developed by following different steps starting from the identification of objectives and aim corresponding to the background evaluation. At the next stage, researchers will get approval corresponding to their selected topic with budget estimations. Researchers then will proceed to search relevant previous research work for developing theoretical concepts by collecting all secondary data. Researchers will gather and manage all the information and selected journals through the development of “inclusion-exclusion criteria”. After evaluation of the literature, they will collect the dataset from “Kaggle ” food processing of the data in “machine learning models”. Then the accuracy of different models will be checked for developing the “fraud detection system”. Then researchers will draft all their findings and evaluations by the following proofreading. In the last stage, researchers will submit the research report based on their findings.
Reference List
Andrews, J., 2023. Credit card fraud. Available at: https://www.money.co.uk/credit-cards/credit-card-fraud [Accessed on: 02-07-2023]
Dastidar, K.G., Granitzer, M. and Siblini, W., 2022, May. The Importance of Future Information in Credit Card Fraud Detection. In International Conference on Artificial Intelligence and Statistics (pp. 10067-10077). PMLR.
Gibbs, S., 2022. Scammers guessed my credit card number – and they could guess yours too. Available at: https://www.theguardian.com/money/2022/feb/26/credit-card-fraud-scammers-guess-attacks [Accessed on: 03-07-2023]
Ileberi, E., Sun, Y. and Wang, Z., 2022. A machine learning based credit card fraud detection using the GA algorithm for feature selection. Journal of Big Data, 9(1), pp.1-17.
Kim, E., Lee, J., Shin, H., Yang, H., Cho, S., Nam, S.K., Song, Y., Yoon, J.A. and Kim, J.I., 2019. Champion-challenger analysis for credit card fraud detection: Hybrid ensemble and deep learning. Expert Systems with Applications, 128, pp.214-224.
Maniraj, S.P., Saini, A., Ahmed, S. and Sarkar, S., 2019. Credit card fraud detection using machine learning and data science. International Journal of Engineering Research, 8(9), pp.110-115.
Najadat, H., Altiti, O., Aqouleh, A.A. and Younes, M., 2020, April. Credit card fraud detection based on machine and deep learning. In 2020 11th International Conference on Information and Communication Systems (ICICS) (pp. 204-208). IEEE.
Sailusha, R., Gnaneswar, V., Ramesh, R. and Rao, G.R., 2020, May. Credit card fraud detection using machine learning. In 2020 4th international conference on intelligent computing and control systems (ICICCS) (pp. 1264-1270). IEEE.
Thennakoon, A., Bhagyani, C., Premadasa, S., Mihiranga, S. and Kuruwitaarachchi, N., 2019, January. Real-time credit card fraud detection using machine learning. In 2019 9th International Conference on Cloud Computing, Data Science & Engineering (Confluence) (pp. 488-493). IEEE.
Trivedi, N.K., Simaiya, S., Lilhore, U.K. and Sharma, S.K., 2020. An efficient credit card fraud detection model based on machine learning methods. International Journal of Advanced Science and Technology, 29(5), pp.3414-3424.
Know more about UniqueSubmission’s other writing services: