Recent technological advancements have fueled a notable increase in credit card usage, consequently amplifying the prevalence of credit card fraud in both offline and online transactions. Although measures such as PIN codes, embedded chips, and supplementary keys like tokens have enhanced credit card security, financial institutions are compelled to bolster their usage controls and deploy real-time monitoring systems to promptly identify and mitigate suspicious activities. This study explores the utilization of ensemble methods, incorporating the k-nearest neighbors (KNN), Random Forest (RF), and Logistic Regression (LR) models, along with the Isolation Forest (iForest) algorithm, to enhance the efficacy of credit card fraud detection. Additionally, automated parameter optimization using GridSearchCV is employed to fine-tune the iForest model parameters. By integrating multiple classifiers into an ensemble approach and automating parameter tuning for the iForest model, our research aims to provide a robust solution capable of adapting to varying datasets and improving fraud detection accuracy. Through empirical analysis and comparison of individual models with the ensemble approach, we underscore the significance of ensemble learning and parameter optimization in enhancing fraud detection capabilities, thereby contributing to the advancement of financial security measures in the realm of credit card transactions.