Carta Compensation

This measures the degree to which our model is capturing the variability in the data points that are used to generate each benchmark. To the extent that our model captures all or most of the variability we would define that as "Very High". To the extent that there is significant variability in the underlying data that requires a heavy degree of modeling to generate a benchmark, we would define that as "Low".

Really though my understanding in simplest terms is that goodness of fit basically says "How well does the target we are showing you thats informed by the actual underlying data points line up with where we thought it was going to be"