Optimizing Credit Fraud Detection, and Analysis Over Imbalanced Datasets

Nowadays, Credit fraud poses a significant threat in an increasingly complex financial landscape, necessitating robust and precise detection methodologies. In today's evolving financial environment, credit fraud constitutes a severe danger that demands rigorous and accurate detection processes. This research employs a strong machine learning approach called K Nearest Neighbors (KNN), Random Forest, XG Boost, and Logistic regression to examine credit fraud detection. Our study tackles the issues of managing skewed datasets, which are typically encountered in fraud detection situations, in addition to proving the usefulness of KNN in recognizing fraudulent transactions. Through rigorous testing and assessment, we acquire essential information on the model's performance measures, such as ROC-AUC scores, accuracy, precision, recall, and F1-Scorel. By leveraging KNN and its competence in handling unequal data, this research greatly increases the toolset of technologies devoted to boosting financial security and inhibiting fraudulent conduct, defending the integrity of financial institutions. This work is useful in enhancing the array of technologies aimed at strengthening financial security and curbing fraudulent activities, thereby safeguarding the integrity of financial institutions. After observation, a hybrid strategy that consist of KNN, and XG Boost is applied to increase the performance of the system.