Development in Machine Learning (ML) Algorithms for Fraudulent Identification in Banking Sector Via Credit Cards

Abstract

The banking industry is seeing an alarming rise in credit fraud, highlighting the critical need for advanced and flexible fraud detection systems. Using state-of-the-art machine learning algorithms, this study systematically reduces the amount of false positives and increases accuracy in identifying credit card fraud. The CCFD dataset will be meticulously prepared to remove any instances of duplicate or missing data. The subsequent stage is to reduce the dimensionality using PCA. To fix the class disparity, the SMOTE method, which stands for Synthetic Minority Oversampling Technique. LR, NB, and XGBoost are three popular older models that are put side by side with the state-of-the-art CCNN. In trials, the CCNN achieved better results than the benchmark models in terms of F1-score (99.97), accuracy (99.96), precision (99.97), and recall (99.97). When compared to XGBoost (97.22%), NB (91.22%), and LR (94.44%), LR produces an exceptional outcome. The model can identify fraudulent transactions, which makes financial transaction systems more secure and dependable. at this exact moment. The XGBoost, CCNN, LR, NB, machine learning, financial sector, credit card fraud detection dataset, and overall financial industry are subjects that are linked to this topic.