Model Improvement Through Transformation of Output Variables

Further improvement of the model can be achieved by transforming the output variables (also called Box-Cox transformation). In this case, functions such as square root, inverse, and logarithm are applied.

This allows a variety of engine based behaviors that are not linear or quadratic in the input variables to be mapped.

To determine the optimal transformation, all transformations are applied to the output and the one with the lowest RMSE is selected. In practice, there is already a lot of experience in applying a specific transformation to certain output variables.