Table 3 |
|||||
The assessment of the cluster dissimilarities | |||||
Cophenetic correlation | Linkage algorithms | ||||
Coefficients (CCC) | Single | Average | Complete | Ward | |
Pairwise - distance algorithms | |||||
Euclidian | 0.69391 | 0.77691 | 0.6625 | 0.57017 | |
Seuclidian | 0.75833 | 0.80679 | 0.64322 | 0.48974 | |
Minkowski | 0.69391 | 0.77691 | 0.6625 | 0.57017 | |
Mahalanobis | 0.75833 | 0.80679 | 0.64322 | 0.48974 | |
Cityblock | 0.72678 | 0.79032 | 0.57675 | 0.54419 | |
Cosine | 0.34928 | 0.81656 | 0.73456 | 0.83168 |
CCC measures the cluster dissimilarities. The most suitable algorithms produce the coefficient which is closer to one “1”. In this case, Ward linkage and Cosine pairwise-distance algorithms generate a coefficient (0.83168) that is the closest to one ‘1’ among other coefficients.
Uragun and Rajan
Uragun and Rajan BMC Neuroscience 2013 14:114 doi:10.1186/1471-2202-14-114