Highly accurate multiple input single output systems (MISO systems) can often be evaluated numerically only with considerable numerical effort. Surrogate models are efficiently evaluable approximations of such MISO systems. The basic challenge here is to construct the surrogate models using as little highly accurate input data as possible. Furthermore, the approximations should be evaluable so efficiently that they allow an almost continuous description of the given MISO system with sufficient accuracy. Surrogate models are used for applications that require a variety of functional evaluations, such as optimization problems or the quantification of design uncertainties.