The method, using a 2-parameter DGBD distribution and a binning method is particularly useful for concave distributions and large datasets (N>100). In particular, it avoids the fit being dominated by low-rank values in the tail.