Published Paper Details:

In current days we can find out lot of loss occurred mainly from credit card transactions, almost billions of dollars are getting lost through credit card fraud transactions. In general to identify such fraud transactions there is no accurate approach and hence the design of efficient fraud detection algorithms is the only solution for reducing these losses and avoids fraud transactions. This problem is very difficult because the credit card fraud data is almost unbalanced and unstructured so there is lot of confusion for the one to classify the fields and then find out the fraud transactions manually. In this thesis we try to design a model using Random forest and CART and take sample dataset collected from KAGGLE website and then check the performance of our model using the algorithms. Here we can see random forest can able to predict and find out the fraud activity very easily and accurately compared with several primitive models. The proposed random forest algorithm achieved more than 99.7 percent accuracy for finding the credit card fraud transactions from a large dataset.