A Logistic Regression-based Model for Identifying Credit Card Fraudulent Transactions

Abstract

The rapid evolution of technology has significantly transformed payment methods, with a notable shift towards online platforms. However, this transition has also witnessed a concurrent increase in fraudulent activities, particularly within online credit card transactions. In response to the escalating occurrences of fraudulent online credit card transactions, this study proposes the development of a robust fraud detection model utilizing machine learning algorithms implemented in Python. Leveraging credit card transaction data sourced from Kaggle, the research utilizes Logistic Regression for both training and testing datasets to identify fraudulent transactions. The efficacy of the model is evaluated using separate test data, resulting in an impressive accuracy rate of 99.87% in detecting previously unseen fraudulent transactions. Further scrutiny of the test data reaffirms this high accuracy, registering a similar rate of 99.8%, thus underscoring the model's adeptness in handling novel data instances. The findings are succinctly represented visually, elucidating the model's efficacy in bolstering online transaction security. By amalgamating advanced machine learning techniques with Python programming, this research contributes to the ongoing efforts aimed at enhancing security measures surrounding online credit card transactions by identifying legit and fraudulent transactions. Such endeavors are paramount in mitigating the adverse impacts of fraudulent activities on both financial stakeholders and consumers.