This PR modifies our slope thresholds by first parsing the positive and negative slope values out and grabbing the percentiles from those split-up ranges. I am hopeful this will make the slope function way less greedy, since it will be identifying more extreme values only.
This PR modifies our slope thresholds by first parsing the positive and negative slope values out and grabbing the percentiles from those split-up ranges. I am hopeful this will make the slope function way less greedy, since it will be identifying more extreme values only.