Detecció de frau en targetes de crèdit/dèbit

Abstract

The importance of fraud detection has increased over the last years, which is why companies, especially banks, are looking for a solution in this regard. This final degree project focuses on the detection of anomalies in banking transactions with the aim of preventing fraud. A methodology based on Machine learning has been applied using data from a real bank. An unsupervised model based on the Isolation Forest algorithm has been developed to identify anomalous patterns in the transaction logs. The project also includes the implementation of the model in the bank, contributing to the improvement of fraud prevention strategies in the bank.