After performing all of the processes in the previous three stages, we now have a well-established data preprocessing pipeline and a correctly trained prediction model. The last stage of a machine learning system involves saving those resulting models from previous stages and deploying them on new data, as well as monitoring the performance, and updating the prediction models regularly.