The influence of experimental designs and the type of model approximation in the formation of metamodels are investigated in this paper. A simple cantilever problem subjected to a bending load was considered for the study. The work used five alternative experimental design approaches and two model approximation types to establish three metamodels for the cantilever problem, namely the volume of the beam, deflection at the free end of the beam, and stress at the fixed end of the beam. The findings of this study demonstrate how the experimental design and model approximation type chosen can have a significant impact on the metamodels correctness. The performance measures for metamodel studied with linear and quadratic approximation on cantilever beam. The work with application of experimental design oriented metamodeling techniques strongly enhances the efficiency of approximation model.