One of the current trends in the aviation world is to work towards an increasingly more computer-aided approach to flying. Despite the improvements, limitations still inevitably exist in terms of power and storage capabilities in the aircraft avionics. To overcome this problem, different solutions have been proposed. A data-driven approach is implemented in this work on a practical application of aircraft performance function. Within the function, the aerodynamics and propulsion submodels are the target of the reduction activity. Neural networks and other surrogate model structures are tested and evaluated on the use case. Notably, several different network architectures are implemented in order to investigate a set of trade-offs between approximation accuracy and model complexity. An analysis of the error introduced by the model approximation is carried out to evaluate the impact at global functional level.