How did other people define their goodness_fit_measure() function? I had trouble with the parameters it should be taking.
It will take as argument the true target and the predictions and compute a sort of distance between them (absolute difference, square difference, etc.)