This is an update on using variations of the MLP method. While I won't go in too much detail (yet) in what each method used, each MLP method had the smoothly falling spectrum and (basically) undershot the monte carlo data. Even though it is not the desired regression, one can still glean something from this. While most of the regressions had a ratio (except for a few bumps) < 1, this tells me that each method suffers this "undershooting" because of some common parameter among them (estimator type, neuron type, training method, etc.). The next step is to start adjusting specific parameters and comparing them with each other while carefully handling the < 5 pT values (to be discussed later). Below you will find regressions for some different methods and ratio plots.

Ratio plots