• Owl [he/him]
    ·
    3 months ago

    Just keep taking derivatives and eventually you'll find one going in the direction you want.

      • Owl [he/him]
        ·
        3 months ago

        Hah, yeah.

        But on a real-world data set, even if the underlying phenomenon is e^x, you'll keep amplifying sample noise until the derivatives are basically random. Assuming you even have enough data to keep taking derivatives.