Using the Gini coefficient to characterize the shape of computational chemistry error distributions

The distribution of errors is a central object in the assesment and benchmarking computational chemistry methods. popular often blind use mean unsigned error as statistic leads to ignore distributions features that impact reliability tested We explore how Gini coefficient offers global representation distribution, but, except for extreme values, does not enable an unambiguous diagnostic. propose relieve ambiguity by applying mode-centered distributions. This version can usefully complement statistics alert on sets with potentially problematic shapes.