Aim: To decrease the error rate of credit card cyber thefts based on binary selection of Novel Random Forest classifier and Linear Regression. The primary scope of this project is to find the malware attacks and it should be informed to the card holder and investigator simultaneously. Materials and Methods: Classification is performed by Random forest classifier(N=34) over Linear Regression (N=34) is for false rate detection. The statistical test difference between g-power 0.08 and alpha values are ????=0.05. Results: The Independent sample T test is applied for the data set fixing confidence interval as 95%. Discussion and Conclusion: Comparison has been made between the Novel Random forest classifier and the Linear Regression for this analysis. The accuracy of the random forest classifier is 94.4% and the Linear Regression accuracy of 51.9%.