Jun 2026· International Journal for Research in Applied Science and Engineering Technology· Vol 14, pp. 2637-2643· 0 citations
TL;DR
This research integrates a bespoke Random Forest classifier with the Synthetic Minority Over-sampling Technique (SMOTE), and results indicate that SMOTE-RF is a feasible and interpretable solution for real-world fraud detection.
Abstract
Credit card theft costs the global banking system billions of dollars every year and erodes user confidence in electronic
payments. Fraudulent transactions are rare (usually < 0.2 % of all records) and this rarity poses a basic problem for automated
detection: a classifier trained on raw data learns to predict the majority class. The resulting model is surprisingly accurate in
general, yet fails to detect much of the fraud it was created for. To directly address such imbalance, this research integrates a
bespoke Random Forest classifier with the Synthetic Minority Over-sampling Technique (SMOTE). The pipeline is tested on the
public European credit card transaction dataset with 284,807 transactions with just 492 (0.172%) fraud cases. The SMOTE-RF
model therefore performs better than Logistic Regression, Decision Trees and the normal Random Forest without oversampling,
with a precision of 0.947, a recall of 0.921, an F1-score of 0.934, a Matthews Correlation Coefficient (MCC) of 0.929 and an
AUC-ROC of 0.983. An ablation research reveals that most of the recall gain is driven by SMOTE, whereas precision is mostly
enhanced by tuning Random Forest hyperparameters. All these results indicate that SMOTE-RF is a feasible and interpretable
solution for real-world fraud detection.
Credit card fraud detection is widely recognized as a critical issue in digital financial systems due to the rapid growth of online transactions. However, existing approaches often fail to provide a comprehensive comparative framework and show limited effectiveness under highly imbalanced conditions. In this context, this study proposes a systematic machine learning pipeline to analyze fraudulent transactions using the IEEE-CIS dataset (590,000 records, 3–4% fraud). The approach integrates data preprocessing, feature encoding, and stratified sampling, followed by the evaluation of multiple supervised models including Logistic Regression, Decision Tree, Random Forest, SVM, Naive Bayes, and XGBoost. Performance is assessed using Accuracy, Precision, Recall, F1-score, and ROC-AUC. The results indicate that ensemble-based methods outperform traditional classifiers. In particular, XGBoost achieves the highest performance (accuracy 98.22%, ROC-AUC 0.9545, F1-score 0.6759), suggesting a strong ability to handle class imbalance. Nevertheless, no model fully optimizes both precision and recall, highlighting a key limitation in current fraud detection systems. Taken together, these findings suggest that ensemble learning provides a robust baseline, while further improvements through hybrid and cost-sensitive approaches are required for effective real-world deployment.
Due to the rapid advancement of digital payment tools, the number of credit card fraud cases is increasing at a faster rate, posing a greater risk to users and financial institutions and highlighting the flaws of traditional rule-based preventive measures. The greatest challenge in this area is that transaction data is extremely unbalanced, with cases of fraud representing only a small portion of the total, and transaction fraudsters continuously develop new methods to overcome specific limits. This paper presents a comparison of popular supervised machine learning models used for detecting credit card fraud. A significant data set of transactions was treated with logistic regression (LR), support vector machines (SVM), random forests (RF), and artificial neural networks (ANN), with imbalance-holding methods, undersampling and oversampling to provide a fair judgment. These findings suggest that the ensemble models, in general, RF, were able to outperform individual classifiers in the sense of higher rates of detecting fraud, with low rates of false positives, resulting in high sensitivity and specificity even in highly biased statistics. Altogether, the discussion shows that ML-driven systems are much more versatile and reliable compared to conventional strategies and provide financial organisations with the means of minimising risks of fraud and protecting payment systems.
Tejaswini H, Vamshi K· International Conference Com...· 0 citations
Credit card fraud is a serious issue in the financial sector and can cause financial losses for customers, banks, and payment service providers. As digital transactions continue to grow, it has become increasingly important to detect fraudulent activity quickly and accurately. Traditional fraud detection systems often rely on predefined rules, which can make it difficult to recognize new or changing patterns of fraud. Machine learning offers another approach by learning patterns from previous transaction data and using those patterns to classify new transactions as legitimate or fraudulent. This research compares four machine learning techniques for credit card fraud detection: Logistic Regression, Random Forest, Support Vector Machine, and XGBoost. The proposed process includes data preprocessing, handling class imbalance, preparing features, training the models, and evaluating their performance. Accuracy, precision, recall, F1-score, and ROC-AUC are considered as evaluation measures. Particular attention is given to precision and recall because fraud datasets are usually highly imbalanced, with fraudulent transactions making up only a small part of all transactions. Overall, simple models such as Logistic Regression can provide an understandable baseline, while ensemble methods such as Random Forest and XGBoost can capture more complicated transaction patterns. The study shows that machine learning can be useful for fraud detection when it is combined with suitable preprocessing, class-imbalance techniques, and careful evaluation.
Jabulani Khumalo, Min Joon Kim· Global Knowledge Academy· 0 citations
The rapid evolution of digital banking and e-commerce puts online payment fraud problems at another level. Because of the large class imbalance present, coupled with the ever-changing nature of fraud, in the instance of credit card fraud, it becomes doubly challenging to achieve real-time detection. Fraudsters continuously evolve and modify their strategies to capture the weaknesses in the digital banking system. To contribute to this problem, this paper provides an approach to detect credit card fraud in the context of machine learning, with a comprehensive set of data preparation methods, the Synthetic Minority Over-sampling Technique (SMOTE) for class imbalance, normalization, feature selection by the information gain, and a hybrid approach using two different supervised learning techniques. The study is based on a readily accessible dataset of European credit card transactions from September 2013, which is composed of 284,807 transactions, of which only 492 (0.172%) are considered to belong to the fraudulent class. In this approach, a Support Vector Machine (SVM) is used to define the critical decision boundaries, and probabilistic outputs from the SVM are used as input to train an artificial neural network (ANN). This hybrid SVM-ANN approach leverages the strengths of both models and improves the classification performance using margin-based learning and nonlinear representation technologies. The study tackles some of the major issues of fraud detection, such as the use of feature sets transformed by PCA for confidentiality, the evaluation of metrics for models that capture rare events, and the generalization of models to be used in imbalanced situations. The proposed approach performs most other traditional classifiers, as evidenced by the experiments and the stated evaluation metrics, which include accuracy, precision, F-measure, and recall. The results show how important it is to use hybrid modeling and strong preprocessing techniques to build scalable and reliable systems for credit card fraud detection.
Firdos R. Kachur, Shivangi Agarwal, Gargi Phadke· International Conference Com...· 0 citations
A comprehensive, real-time credit card fraud detection framework that integrates geospatial and occupational user behavior features with supervised machine learning models is proposed and successfully bridges the gap between theoretical machine learning models and practical, low-latency deployment in financial technologies.
Narayana Reddy T, Arun Kumar T M, Nagaraju V et al.· American Journal of AI Cyber...· 0 citations
A UPI Fraud Detection System using the Random Forest machine learning algorithm that helps banks and digital payment platforms identify suspicious transactions quickly, reducing financial losses and improving user security.
Avu Siva Dinesh Kumar, G. Kumar· International Journal for Re...· 0 citations