Physical and statistical models are combined to describe and design magnesium and high entropy alloys. A principal component analysis is applied to merge material datasets, and it is shown that limits in properties can be envisaged. Extrapolation techniques can be employed to devise properties of non-existing alloys, such as specific heat capacity, melting point and Young’s modulus. These in turn can be input to physical models to predict, for example, yield strength and modulus of toughness. The tools described herein can readily be used for materials discovery, and are being implemented in the Accelerated Metallurgy project.