An Optimized Credit Card Fraud Detection Using Balancing Techniques: A Comparative Machine Learning Study Sampling & Testing Integrated

Credit card fraud is a massive problem for financial institutions and consumers due to the intrinsic imbalance between legitimate and fraudulent transactions in data sets. The implementation of advanced machine learning algorithms is required to enhance detection accuracy and minimize false positives since conventional detection methods often yield high false negatives and overfitting. To overcome these barriers, the Optimized Credit Card Fraud Detection system proposed here utilizes advanced machine learning methods. To counter dataset imbalances and enhance detection rates while reducing false positives and negatives, we explore and compare various testing and sampling strategies. Our research evaluates the performance of advanced balancing techniques such as Edited Nearest Neighbours (ENN), Tomek Links, Adaptive Synthetic Sampling (ADASYN), and Synthetic Minority Over- sampling Technique (SMOTE). Our research evaluates model effectiveness like that of Long Short-Term Memory (LSTM) networks, Random Forest, Support Vector Machine (SVM), and Logistic Regression. For further enhancing the detection abilities, we also employ emerging deep learning techniques such as Generative Adversarial Networks (GAN) and Convolutional Neural Networks (CNN). In order to identify a sound fraud detection approach, models are analyzed using precision, recall, F1 Score, and AUC-ROC metrics. Our findings demonstrate that integrating these state-of-the- art approaches significantly enhances overall precision and reduces false positives. Through the utilization of balanced datasets and state-of-the-art AI methods, this paper presents a tailored solution for credit card fraud detection with higher efficacy and efficiency, leading to an improved and streamlined detection system. Implementation of new algorithm called Fraud Sight Boost gives more precision and accuracy in CreditCard Fraud Detection By setting a new benchmark for fraud detection systems, the proposed framework invites research and realworld application.