Point-Of-Sale (POS) system has become ubiquitous and popular among micro and small-scale businesses such as retail stores, supermarkets, and other businesses for daily transactions especially in Nigeria. Among its numerous advantages such as better inventory management, simple invoicing, quick payment and others, it is fraught with a lot of security challenges or attacks some of which include: malware attacks, key logger attacks, and user identity attacks. The future of this promising technology looks bleak if these breaches or attacks are not identified and checked. This research proposes a detailed novel security architectural design identifying areas of possible breaches and possible solutions. It utilized KMeans clustering and KNearest Neighbour (KNN) algorithms on data collected from POS to classify the data generated and achieved an impressive result of 58.17% clustering separation and 99.51% accuracy classification of data points respectively.