Hi Laura,

If you have the ground truth available for the dataset(s) in question, you can use NMI or ARI to compare the clusterings at an appropriate level of the returned dendogram.

SOM and HAC (Hierarchical Agglomerative Clustering) methods are not directly related. HAC methods return a dendogram while SOM yields a clustering. However, you can think of the clustering returned by SOM as being analogous to the clustering at the appropriate level of the HAC dendogram.

Hi Laura,

Computational cost, time, and accuracy are the metrics normally used for clustering algorithms.

Hi Laura,

power consumption, No of dead nodes per round, accuracy, and nodes position distribution i think.