Classification of Customer Actions on Digital Money Transactions on PaySim Mobile Money Simulator using Probabilistic Neural Network (PNN) Algorithm

Abstract

Development of technology have influenced all aspect, especially in financial sector in this pandemic situation, where most people tend to use digital money to conduct daily financial transactions. In one side, there is security point that need to be concern much. Like several disadvantages using credit cards by undue owners, social engineering, and transactions to commit fraud. In this paper, PaySim Mobile Money Simulator data is used with a machine learning algorithm called probabilistic neural network (PNN) to classify whether the customer's actions are normal or fraudulent actions. This PNN approach combined using binary classification to prevent fraudulent actions in transactions that have been or are being used by customers. And the result indicated that this system able to classify class 0 (as a normal class customer) and 1 (as a fraudulent class customer). Based on this result, maybe it would help many sectors that involved as a tool to classify a genuine customer. Especially in this pandemic covid-19, the fraud needs to detect often, to mitigate the fraud early.