I’d like to extract the necessary information from the Hierarchical Cluster View node to enable me to order the columns of a table in the same way as they appear in the tree. This is easy to do in R, but I can’t see any way to do it with Knime’s clustering tools. I’d rather not have to call up R just to do something as simple as this. Is there any way to do it?
The Hierarchical Cluster View node doesn’t have an output port, so you can’t use it to pass information to other nodes. Just to confirm, you want to re-order the columns in your data to reflect the ordering of the clusters?
Yep, I’d like to order the rows or columns of a table so that they match the left-to-right order of the items in the dendrogram. A workaround that gets me half-way there is to assign clusters at multiple levels (e.g. once with 10 clusters and once with 20) and then group the smaller clusters within the higher-level clusters. But matching the dendrogram exactly would be preferable.
I think my extension had similar feature, the Sort by Cluster node. (Also there was a Leaf Ordering node, which order the results to minimize the sum of adjacent diffs. Reverse Order allows to change the order to the opposite. There is also a Dendrogram with Heatmap node if you prefer that visualization.)