We develop a Markov chain approach to forecast the production output of a human-machine system, while encompassing the effects of operator learning. This approach captures two possible effects of learning: increased production rate and reduced downtime due to human error. In the proposed Markov...