Does WEKA Cluster Assigner work correctly?

Dear KNIME developers,

I tried to use the WEKA DBScan Node to cluster some data. Afterwards I wanted to assign the determined clusters to the same data using the WEKA Cluster Assigner. Unfortunately the assigned clusters differ from the DBScan results although I'm using the same data. Furthermore when I have a look at the output of the Weka Cluster Assigner the cluster membership columns aren't consistent with the winner column.

So did I use the nodes in a wrong manner or is this a bug?

I'm using the KNIME Version 2.7.1 and the WEKA 3.6 integration.

Thanks in advance for your answer.

Kind regards!

This his hard to explain, I guess we would need to have the workflow or some way to reproduce it...

I have for example tried to cluster the Iris data with DBSCAN. I've attached a screenshot of the Workflow, where you also can see my choice of eps and minPts for DBSCAN. Furthermore please find attached a screenshot of the results.

I don't understand why especially the last rows get a Winner Cluster although they aren't assigned to any of the three clusters if you have a look a the bars. Furthermore if you have a look at Row82 then you see that it doesn't have a winner cluster in spite of being fully assigned to the first cluster. I you have a look at the WEKA node view (screenshot attached), than you can see another different result for Row82.

These are the things that I don't understand and which made me suppose that the nodes don't work correctly. But maybe I don't use them in the right manner!? Thank you for your help!