Why clustering with k-means?

[richardheery]

Hello,

I am wondering if the use of k-means in the parameters km, row_km and col_km may be a misnomer? It seems to me that the resulting clusters are simply the result of cutting the dendrograms produced by hierarchical clustering associated with the rows and columns to achieve the required number of clusters. Using K-means clustering could give quite different clusters to those produced using hierarchical clustering and so showing them on the dendogram would be quite confusing.

Or did I miss something?

Thanks

Richard

[Yunuuuu]

You can use row_split and column_split to split by dendrogram