Algoritmo de estimación del porcentaje de fraudes en transacciones con tarjeta de crédito

Abstract

One of the main problems that arise in companies is fraud due to electronic transactions, generating financial losses for these entities. Banks have a whole security network that makes it possible for them to detect this type of fraud, but small companies do not have the opportunity to quantify this type of fraud to minimize the losses that this generates. The project proposes the execution of an algorithm that allows the detection of a specific percentage of users suspected of having committed fraud in electronic transactions with credit cards, taking into account that the development is based on statistical methodologies. It is sought that the result of the algorithm can be of great help for companies to take measures according to the results obtained. The algorithm plans to give a specific number of credit card frauds from an experimental database where the raised data is encrypted with data mining techniques such as PCA (Principal Component Analysis) transformation in order to protect the information since the entities do not provide data on their customers for security reasons. The technique provided seeks to help small and large companies to detect in an automated, easy and safe way; numbers of frauds that may be causing losses in their companies.