This paper proposes a federated learning model for credit card fraud detection (CCFD) with data balancing techniques. The study addresses two main challenges: data privacy and class imbalance in credit card transactions. Due to data security and privacy concerns, banks typically do not share their transaction datasets, making it difficult for traditional systems to learn and detect fraud. To overcome this, the paper introduces a federated learning approach that allows different banks to collaborate without sharing raw data, preserving data privacy. The study also tackles the issue of class imbalance, where fraudulent transactions are rare compared to legitimate ones. To address this, the paper evaluates various resampling techniques, including oversampling (e.g., SMOTE, ROS, AdaSyn) and undersampling (e.g., RUS), and compares their effectiveness with different classification models. The results show that hybrid resampling methods outperform individual methods in terms of accuracy, recall, precision, and F1-score for machine learning classifiers such as Random Forest (RF), K-Nearest Neighbors (KNN), and Decision Tree (DT). The best accuracy values for these classifiers are 99.99%, 99.96%, and 99.98%, respectively. The proposed federated learning model is implemented using two frameworks: TensorFlow Federated and PyTorch. The PyTorch framework achieves higher prediction accuracy for the federated learning model but requires more computational time compared to TensorFlow Federated. The study also evaluates the performance of the model on different platforms, including PyTorch and TensorFlow Federated, to determine the best platform for accuracy and computational efficiency. The paper concludes that the proposed hybrid resampling techniques and federated learning model effectively address the challenges of data privacy and class imbalance in credit card fraud detection. The results demonstrate that the model achieves better performance than traditional methods, making it a promising solution for detecting credit card fraud in a secure and efficient manner.This paper proposes a federated learning model for credit card fraud detection (CCFD) with data balancing techniques. The study addresses two main challenges: data privacy and class imbalance in credit card transactions. Due to data security and privacy concerns, banks typically do not share their transaction datasets, making it difficult for traditional systems to learn and detect fraud. To overcome this, the paper introduces a federated learning approach that allows different banks to collaborate without sharing raw data, preserving data privacy. The study also tackles the issue of class imbalance, where fraudulent transactions are rare compared to legitimate ones. To address this, the paper evaluates various resampling techniques, including oversampling (e.g., SMOTE, ROS, AdaSyn) and undersampling (e.g., RUS), and compares their effectiveness with different classification models. The results show that hybrid resampling methods outperform individual methods in terms of accuracy, recall, precision, and F1-score for machine learning classifiers such as Random Forest (RF), K-Nearest Neighbors (KNN), and Decision Tree (DT). The best accuracy values for these classifiers are 99.99%, 99.96%, and 99.98%, respectively. The proposed federated learning model is implemented using two frameworks: TensorFlow Federated and PyTorch. The PyTorch framework achieves higher prediction accuracy for the federated learning model but requires more computational time compared to TensorFlow Federated. The study also evaluates the performance of the model on different platforms, including PyTorch and TensorFlow Federated, to determine the best platform for accuracy and computational efficiency. The paper concludes that the proposed hybrid resampling techniques and federated learning model effectively address the challenges of data privacy and class imbalance in credit card fraud detection. The results demonstrate that the model achieves better performance than traditional methods, making it a promising solution for detecting credit card fraud in a secure and efficient manner.