A machine learning apparatus, which learns a condition associated with arrangement positions of magnets on a core of a motor rotor, includes: a state observation unit that observes a state variable composed of data relating to arrangement positions of a plurality of the magnets arranged on the core, data relating to a magnitude of cogging torque generated when the plurality of the magnets are arranged at the respective arrangement positions, and data relating to a magnetic flux density of each of the magnets; and a learning unit that learns the condition associated with the arrangement positions of the magnets on the core in accordance with a training data set constituted by the state variable.