ABSTRACT

The project seeks to investigate the influence of data structure and algorithms in the field of financial technology (Fintech), with a special focus on using the Random Forest algorithm to categorise the credit card dataset. The Random Forest technique is used to train individual trees to build decision trees, which are efficient and dependable for handling big data sets. It is also used to assess the accuracy of fraud detection in transactions. The study used range partitioning, Sliding-Window, step-2, and independent classifiers to extract fraudulent traits from the cardholders' behaviour. Nevertheless, the imbalanced nature of the dataset resulted in challenges for classifiers to achieve satisfactory performance. Two primary methodologies were employed: one-class classifiers on the original dataset. Ultimately, all individuals in the group received training utilising the same classifier, with the most esteemed classifier being the most recent behavioural pattern for each cardholder. The research effectively showcased the use of the Random Forest algorithm for credit card fraud detection. It efficiently categorised transactions and accurately identified fraudulent activity with a high level of recall and accuracy. By using machine learning techniques and advanced data analytics, this study aims to enhance the ability of financial institutions to identify and prevent fraudulent transactions, hence safeguarding the integrity of the financial system. Ultimately, the project emphasises the significance of comprehending the impact of data structure and algorithms in the field of financial technology. It offers valuable guidance for developing and enhancing efficient algorithms, as well as promoting cooperation among regulatory agencies, financial institutions, and industry professionals.