This paper presents a rule-based machine learning model (RBM) for financial fraud detection, aiming to address the challenges posed by highly imbalanced datasets and the evolving tactics of fraudsters. The RBM does not rely on data resampling techniques, making it more interpretable and explainable compared to other machine learning models. The effectiveness of the RBM is evaluated using various metrics such as accuracy, precision, recall, confusion matrix, Matthew's correlation coefficient (MCC), and receiver operating characteristic (ROC) values. The model is compared with several existing machine learning models, including random forest (RF), decision tree (DT), multi-layer perceptron (MLP), k-nearest neighbor (KNN), naïve Bayes (NB), and logistic regression (LR) using two benchmark datasets: PaySim and BankSim. The experimental results show that the proposed RBM outperforms the other models, achieving high accuracy and precision of 0.99 and 0.99, respectively. The RBM's performance is further validated through ROC curves, confusion matrices, and rule generation and validation processes. The study concludes that the RBM is a robust and efficient tool for detecting fraudulent financial transactions, offering transparency and interpretability in the learning process, which is crucial for the financial sector. Future work will focus on reducing the rule generation and classification process time to further enhance the model's performance.This paper presents a rule-based machine learning model (RBM) for financial fraud detection, aiming to address the challenges posed by highly imbalanced datasets and the evolving tactics of fraudsters. The RBM does not rely on data resampling techniques, making it more interpretable and explainable compared to other machine learning models. The effectiveness of the RBM is evaluated using various metrics such as accuracy, precision, recall, confusion matrix, Matthew's correlation coefficient (MCC), and receiver operating characteristic (ROC) values. The model is compared with several existing machine learning models, including random forest (RF), decision tree (DT), multi-layer perceptron (MLP), k-nearest neighbor (KNN), naïve Bayes (NB), and logistic regression (LR) using two benchmark datasets: PaySim and BankSim. The experimental results show that the proposed RBM outperforms the other models, achieving high accuracy and precision of 0.99 and 0.99, respectively. The RBM's performance is further validated through ROC curves, confusion matrices, and rule generation and validation processes. The study concludes that the RBM is a robust and efficient tool for detecting fraudulent financial transactions, offering transparency and interpretability in the learning process, which is crucial for the financial sector. Future work will focus on reducing the rule generation and classification process time to further enhance the model's performance.