Aim: The main aim of the research work is to create and build a novel fraud detection approach for Streaming Transaction Data, with the goal of analyzing historical customer transaction details and extracting behavioural patterns. Cardholders are divided into groups based on the volume of their transactions. Materials and Methods: The categorizing is performed by adopting a sample size of n = 10 in K-Nearest Neighbour and sample size n = 10 in Gaussian Naive Bayes algorithms with G power in 80% and threshold 0.05%, CI 95% mean and standard deviation . For the implementation, the FraudTest dataset was used. Results : The analysis of the results shows that the K-Nearest Neighbor has a high accuracy of (99.53) in comparison with Gaussian Naive Bayes algorithm (81.95). There is a statistically significant difference between the two groups with value p=0.005 (p< 0.05). Conclusion: The results show that the K-Nearest Neighbor algorithm for detecting fraud in credit card transactions appears to generate better accuracy than Gaussian Naive Bayes algorithm.