Fig. 5
From: A new fast method for inferring multiple consensus trees using k-medoids
Comparison of our algorithm (â—‡) based on the k-medoids clustering, the non-squared RF distance and the SH cluster validity index to the traditional approach (â–³) based on the k-means clustering, on the squared RF distance and on the recomputing the majority consensus trees within k-means (Stockham et al. [17]). The coalescence rate parameter in the HybridSim program was fixed to 5 in this simulation. The comparison was made in terms of ARI (panels a and b) and the running time (measured in seconds) of the methods (panels c and d) with respect to the number of tree leaves and trees