Figure 3

Comparison of VIS clustering among data sets. We developed two methods to describe the extent of VIS clustering. The first method 'maximum %' is simply the maximum bin's z-score divided by the total number of VIS in the data set, $100\cdot max\left(X\right)∕\sum _{i=1}^n{C}_i$. Data sets with a maximum % > 8 indicate a high degree of clustering. The second method 'BCP posterior probability' is calculated after running the Bayesian change-point analysis, and is simply one minus the average of the posterior probabilities of a change point occurring at each bin, $1-\overline{P}$. BCP posterior probabilities > 0.98 indicate a high degree of clustering. Both methods indicate that the CGD data exhibits a high degree of clustering with a maximum % and BCP posterior probabilities of 11.98 and 0.999, respectively, in comparison to the other data sets which ranged from 0.5-1.48 and 0.9356-0.9361, respectively.