variable labels and value labels


does anybody know if is it possible to attach variable labels and value labels to data in KNIME, like it can be done with spss or sas?

Thanks, Kees

Hi Kees,

any chance you can share what you are trying to achieve that would require "variable labels" and "value labels"?

While KNIME may not have something like that right out-of-the-box, chances are there coluld be many other ways to achieve the same end-result using some of the available features.

For example, if you need to doccument part of your workflow you can use Workflow Annotations. If you need to document specific nodes more in details, you can use Node Descriptions and so on.

Also take into account that KNIME data table columns can have descriptive names including spaces, special characters and more which can be referenced later on in graphs, reports, etc. 

Hope this helps.


Hi Marco,

Sorry for not explaining better in the first place.

Variable labels are annotations for columns, so as to better describe the meaning of the column. 

Perhaps I should say column or attribute labels, since a variable would indeed have a different meaning in KNIME.

Value labels are annotations for different values of the columns. For the same reason.

For instance, variable label for numeric column 'satisqual'  with possible values 1, 2, 3, 4, 5  could be 'satisfaction with the quality of the product' having variable labels 1='very dissatisfied', 2='dissatisfied', 3='neutral', 4='satisfied', 5='very satisfied'. Then you could make calculations on the numbers and see the labels in the output, for instance with frequencies or crosstabulation. That is the idea. 


In SPSS you can label values of numeric as wel as well as string columns.

Other reasons for labeling numeric values instead of using string variables had to do with numbers being more economical in Kb.

With R this is an example:

Thanks, Kees

Thanks Geo. I see your point. 

I know what you refer to. This idea does not exist in KNIME. And it is also not really needed.

For variable labels, you can already define long column names in KNIME with any character except $. In the worst case, create a second table with Table Generator with variable names and their corresponding labels, which you can join using Joiner when needed. You'd have to Unpivot the variables for which you would attach the variable lablels, e.g. for visualisations. The advantage of KNIME is that you can manage several tables at the same time.The same idea can be used for value labels, if you need them.

After all, I guess that's what SPSS or Stata would be doing behind the scenes, i.e. secretly refer to a another invisible table / list so that you can enjoy the magic of value and variable labels :-)

Geo. your approach worked perfectly for me. Thank you for explaining the rationale and the solution. This is no doubt what SPSS and Stata do internally, and it’s very easy to do.

1 Like