GroupBy doesnt work with flow variables anymore

Hi,
I was using the GroupBy node with flow variables before. I generated a list of columns names that I wanted to iterate over and then used the TableRowToVariableLoopStart to calculate the mean and standard deviation for each iteration for one column (yes, there is a reason why I do it like this and not calculate all columns at the same time). The same was also possible using the ColumnListLoopStart node.


Anyway the GroupBy node needed to be configured like this:
- add 2 aggregation columns in options (can be any) and select mean and sd as methods
- go to Flow Variables tab and select in aggregationColumns -> columnNames -> there are two entries available now: s0 and s1
- select the flow variable for s0 and s1 and then node told me that these are now under control of a flow variable.

Now, this worked in the past (1-2 years ago) and it doesn't any more (already since few months), instead I get these messages:
WARN      GroupBy                            No aggregation column defined
WARN      GroupBy                            2 invalid aggregation column(s) found.
… and no mean and sd are calculated. I copy/pasted the groupBy node from an old workflow (2 years old) and configured it in the same way and here it works all fine.
I guess something has changed or a bug was introduced….
Need more information or an example workflow?


Thanks,
Martin

Hi Martin,

which version of KNIME do you use? A simillar bug was reported for version 2.9.0/2.9.1 and was fixed with version 2.9.3.

Sorry about that.

Tobias

Hi Tobias,

I was working with 2.9.2 already, now I updated to 2.9.4 and cannot reproduce the error. So I guess it had been fixed with some update.

Problem solved, thanks!

Martin

Hi Martin,

you are welcome :-) I'm glad it works now. We fixed the bug 2.9.3. not in 2.9.2. I just updated my former post to reflect that.

Bye,

Tobias

 

Hi Tobias,

could it be that this bug is back? Since 2.10 I have again the same problem...

I have a strange behaviour, now:

- in 2.10 and 2.11.2 I successfully run 2 out of 5 iterations and the last three fail with the error message:

WARN      GroupBy                            No aggregation column defined

WARN      GroupBy                            2 invalid aggregation column(s) found

Failing mean it runs through, but the GroupBy doesnt aggregate any values.

- however, when I try to run the Iris data set in the example workflow, all four iterations run normally!??? 

To summarize, the process I try to run is just Column List Loop Start followed by GrouBy with Aggregation columns controled by the flow variable currentColumnName for a mean and a SD aggregation ... and Loop End. 

Cheers,

Martin

Hi Martin,

sorry to hear that. I am not aware of any changes that might have caused this problem. Would it be possible for you to send me the workflow created with KNIME 2.9.4 to contact@knime.com. If not with the original data maybe with some dummy data. This would make it much easier for me to locate the problem.

Thanks,

Tobias

Hi,

in general, if someone is using flow variables to change the aggregation columns I would suggest using the new  Pattern Based Aggregation column selectors (patternAggregators in the Flow Variables tab). These allow you to specify an aggregation method for all columns which names match a given pattern e.g. the column name. See the attached workflow as an example.

Bye,

Tobias

A post was split to a new topic: Problem with Pivoting Node