Publicación:

Mapeo de las técnicas de aprendizaje automático usados en fraudes con tarjeta de crédito

Resumen

Fraud is a criminal act, which is taken advantage of by people who see the opportunity to do it and execute these facts against the finances of the companies. This crime is present in almost all organizations and losses represent around 5% of income per year, its early detection or failing in real time is a pressing need and requires effective solutions that preserve prestige and reliability for customers. Therefore, this study analyzes the machine learning techniques used in credit card fraud. Methodologically, it is a qualitative descriptive-explanatory study, addressing the techniques commonly used in the Automatic Learning processes in the typology of credit card fraud. The result of this research reflects that there are many investigations that have been carried out on the subject, covering great advances since 2002. It is observed that learning algorithms such as neural networks, logistic regression and decision trees have contributed to give effective solution to the Timely detection of even credit card fraud.