The need for an energy efficient and high-performance traction system has been rapidly increasing over the years which has driven the interest in designing electric motors for electric vehicles (EV’s). Among different motor types, the use of Interior Permanent Magnet (IPM) motors has been dominant because of the high torque characteristics and, most importantly, reliability. Anyway, because of the multi-physics nature of the design of motor which deals with thermal, electromagnetic and structural considerations, it becomes computationally expensive for the optimization problem when completely relying on the Finite Element Analysis (FEA) alone. This thesis is built upon a data-driven strategy to improve the design and optimization of traction electrical motors by integrating high-quality simulations with machine learning techniques.