Download the following dataset and set it to working directory. In this project we are trying to detect frauds in credit card details given by customers to bank. We are applying two types of models Supervised and Unsupervised learning. In Supervised Learning Model Kernel Principal Component Analysis integrated with Logistic Regression makes a good classification model. it gives accuracy approx 97%. other method in supervised learning is Artificial neural network. it gives accuracy approx 97.54%. And here comes a very special method in Unsupervised learning Self Orgnising Feature Map integrated with Artificial Neural Network . it is a hybrid model of unsupervised and supervised learning. and it gives accuracy of 99.24% with min loss. and that's hugh.