Has to be verified, but in my investigations a 3rd degree polynomial captures everything interesting and is robust to missing segments. np.polynomial.polyfit is also very fast.
Still, it will add another layer before data are saved, I think most users would not want that. Although one could argue that what is saved already comes from an interpolation?
Has to be verified, but in my investigations a 3rd degree polynomial captures everything interesting and is robust to missing segments. np.polynomial.polyfit is also very fast.