Gap Statistic Across Varying Cluster Sizes for HGDP Africans and NECS. The graph plots the mean gap statistic (see Methods) at each cluster size. The gap statistic for the HGDP Africans steadily increases to 21 clusters at which point the statistic cannot be computed due to singleton clusters. For the NECS, the statistic again increases with an increasing cluster size and does not reach a point after which the statistic continuously decreases. In both situations choosing an optimal cluster size is not obvious.
Solovieff et al. BMC Genetics 2010 11:108 doi:10.1186/1471-2156-11-108