Bank transaction Fraud is a deliberate act of deceit involving financial transactions to obtain a personal advantage. As the quantity of online transactions has risen, so has the number of scams. Detecting fraud is critical in the banking business to safeguard clients' funds, minimize fraud losses, and maintain profitability. Banks are using machine learning-based models because traditional fraud detection approaches are no longer adequate for identifying fraud. Skewness is a significant issue with financial transaction data, and any model's performance is data-dependent and technique- dependent. Using multiple parameters, this article compared several machine learning models Distributed Test Checkpointing approach. The research examined mobile money transactions reported on Kaggle during the previous six months. Python was used to develop the machine learning Distributed Test Checkpointing approach, while Sk learn and pandas were used to analyze the data. Analyses after the random forest, SVM with proposed novel checkpoint-based approach perform better than the other models.