Data mining techniques for proactive fault diagnostics of Electronic Gaming Machines

Abstract

This paper details the preliminary research into modeling the behavior of Electronic Gaming Machines (EGM) for the task of proactive fault diagnostics. The EGMs operate within a state space and therefore their behavior was modeled, using supervised learning, as the frequency at which a given machine is operating in a particular state. The results indicated that EGMs did exhibit measurably different behavior when they were about to experience a fault and these relationships were modeled effectively by several algorithms. © 2010 Springer-Verlag Berlin Heidelberg.