is there any node or simple approach to get the relationship (1:1, 1: many, many:many) between columns?
It would help to check big tables. So far I tried to check it with a lot of GroupBy nodes.
This leads me to another suggestion for the GroupBy node: It would be nice to have an option (checkbox?) to get a simple rowcount per group which does not require the selection of a non-grouping-column (I even had the case where I had to create a dummy column as I needed to group on all columns and still wanted the row count per group).
how do you mean relationship? Like a mapping of the values?
Maybe the linear correlation node. It gives you a correlation matrix. 1:1 should be a correlation of 1.
For the groupby we already have an open feature request.
I'm looking for database like releationship. I would (for example) like to have a simple check if each (unique) entry in one column has exactly one (unique) entry in the other column and vice versa (would be a 1:1 releation). I can group by those columns and count for each unique entry in one column the unique entries in another columns. But that is just one direction and finally I do not care about the count for a single unique entry. I'm only interested if I have count one for all unique entries (which requires another grouping).
Could I explain the problem?
It is not that I cannot solve it in KNIME but I need it often and it takes several nodes to get to that result. I was wondering if there is an easier way.