Association Rule Learning, empty output but Association Rule Learning Borgelt works

I have a KNIME order basket analysis and obviously try to find patterns. I prepare the dataset with an ETL style workflow which is collapsed into a metanode. With the prepared dataset , an association rule learning borgelt ist being executed and works - well it produces an output table according to the given input/confidence parameters and the result looks reasonable.

with the very same Input, an assocation rule knode is being executed too but produces an empty output.
I have no clue why. The question behind could be: how to debug the execution of nodes ?

The workflow looks like:

the ETL output table is:

the number input knode configuration is:

the config of the assocation rule learning (borgelt) knode is:

the output of the association rule learning (borgelt) knode is:

the config of the association rule learning knode is:

but no output, warning in the console:

any ideas would be appreciated, thank you.

Hi @marcop -

I loaded up a similar workflow from the EXAMPLES server to check on this. I think the reason the Association Rule Learner Node is returning an empty table is that the minimum support is too high. Can you lower it significantly - try something like 0.05 - and see if you you get any results?

Something else to note is that you are specifying a percentage for the support in the Borgelt learner, so I think you are probably not making an apples-to-apples comparison with the Association Rule learner.

1 Like

Dear ScottF

thank you for your reply and the example workflow, I’ll check it.
However, lowering the minimum support did not produce any output. I’m happy that the Borgelt node is runnig.

regards

worked for my dataset! much thanks!

1 Like