In short, my initial df has a column that has probabilities from an external predictive model that I would like to compare to the predictions generated from my lightGBM mode