Hi,
why if you group for some dimension, the very same disappear from the aggregation menu?
Quite often i have to group for something and then count the number of row.
In alteryx you can group and compute some aggregation with the same field…
Luca
I agree with @Luca_Italy that this is inconvenient. KNIME mostly works as a visual language not an end user tool. So, to add column or variable you need to use a visual code (node).
Hi @Luca_Italy and @izaychik63,
I kind of feel I agree because I found the GroupBy awkward to use for similar reasons recently, but in putting together an example workflow I’m wondering to what extent it is a problem, and how much it just feels like it’s a problem.
Attached is a demo simple table of item codes, values and age as in the attached workflow.
I then group on item code,
I can get, for example, a sum of value, and max and min of age. I want to also get a count of item_code too (which as you say I cannot choose), but I can still get the count by counting either value or age. It still gives the total for the group even though it does get a less than friendly name like count(value) which isn’t so good. (I’d probably want to rename the column).
KNIME_groupby.knwf (8.3 KB)
Does this maybe become more of an issue if you want to group by two columns and perhaps then try to get counts at both the main group and subgroup?
Using the attached workflow, or a more complex variation on it, what is it that you might want to be able to do, but can’t? I’m interested myself, as it’s useful to discover limitations that I haven’t considered, and maybe there are also some useful workarounds for problematic use cases.
