Establishing systems that can detect digital transactional fraud is essential because the in the past developments in e-commerce sites and e-payment-gateways are leading to an increase in finance fraud cases. The selection of features related to digital transactional fraud especially in credit card scams is critical in machine learning applications for fraudulent detection and should be chosen with care. There is a prevailing belief that credit card transaction fraud poses a growing threat with significant repercussions for the financial industry. In this context, data mining emerges as a pivotal tool in detecting credit card fraud across both on line and offline transactions. The efficacy of fraud detection in credit card transactions is notably influenced by the dataset’s sampling method, the selection of variables, and the detection techniques employed. This study delves into assessing the performance of Decision-Tree(DT), Random Forest(RF), Logistic-Regression(LR), K-Nearest-Neighbors(KNN), and Support-Vector-Classifier(SVC) algorithms on credit card fraud data. The dataset, comprising 284,807 transactions from European cardholders, underwent a blend of under-sampling and oversampling techniques to address data imbalances. Five strategies were applied to both raw and preprocessed data sets, respectively. This study underscores the pivotal role of data mining techniques in fortifying credit card transaction security, exemplifying the imperative nature of proactive measures in safeguarding digital transactions, as exemplified by SecureSwipe.