What are the drawbacks of Normalized Mutual Information (NMI) clustering evaluation method? For evaluating what clustering algorithms, is the NMI evaluation method suitable?
It is a matter of preference.
If you come with an information theory background, you'll like NMI better. If you come from a classification background, Rand index is pretty much the "accuracy" that you are used to, but on pairs.
In Comparing Community Detection, whenever the ratio between the number of members and the number of clusters is small the Normalized Mutual Information becomes too high which is called selection bias problem. This problem caused by inclination of selecting solutions with more clusters (Amelio et al. 2015)
For evaluating what clustering algorithms, is the NMI evaluation method suitable?
I think this question incomplete.