A Genetic Algarithm For Fraud Detectian E/ectranic Debit Cards In Peru

Abstract

The great problem of the bank is to detect fraudulent transactions which are scattered with genuine transactions. The proposed solutions are not sufficient to detect these illegal operations precisely because they are aimed at different Peruvian markets. Consequently, we reviewed the main con­ tributions in this area, such as Dempster-Shafer theory and Bayesian learning, Fuzzy Darwinian, etc. presents the modeling and design of a genetic algorithm to obtain more representative rules cardholder purchase within the universe of transactional data collected from a Peruvian bank. From experimental evidence obtained an accuracy of 95.5% in the interne! channel 95.8% for the point of sale. Finally, it was concluded that the use of evolutionary algorithms strategy gol an acceptable accuracy in prediction.