Example of results on a few datasets.

The table represents the results of the three clustering measures (QLC, TE and HS) over 10 NJ reconstructed methods. For the measure QLC higher scores mean better performance, whereas for the measures TE and HS lower values mean better performance. It can be seen that in some cases, for example dataset #7, all the distance measures (except for Euclidian) give the same best score (1 for QLC and 0 for TE and HS). For dataset #1, on the other hand, all the distance measures give a rather similar bad score. It is not surprising that the scores of a dataset which contains 9 individuals (#1) will be lower than the scores of a dataset which contains only 2 individuals (#7). There are cases, like #3 and #4, in which the best performance is achieved by one method (NJ-Normalized Absolute) and other cases, like #5, where a few methods receive a very high score (NJ-Normalized Absolute, NJ-MMM length dependent rates, and more).