The diagram presents two stages. The first stage is training of a surrogate model. Input data x subscript i flows into a mathematical or numerical model. This produces actual output y subscript i. Both input data x subscript i and actual output y subscript i feed into supervised training. The supervised training process generates a surrogate model. The second stage is application of the surrogate model. Input data x subscript i is passed into the surrogate model. This produces predicted output y hat subscript i.