Difference between Dropdown and Text Input in Flow Variables Tab?

KNIME Analytics Platform
1

Hi,

what is the difference between the dropdown box and the text input field in the "Flow Variables" tab? Please forgive me if I miss the obvious, but I was not able to figure this out...

I tried to implement parameter variation on the random forest learner in a loop. To do so e.g. for the nrModels property, I select my flow variable from the dropdown box and everything seems to work fine.

If I try to vary the splitCriterion property by again selecting my corresponding flow variable in the dropdown box, I get «Errors loading flow variables into node : Unable to parse split criterion "Information Gain"»

A screenshot in a post dating from 2014 (https://tech.knime.org/forum/knime-general/parameter-optimization-and-dialog-options-in-learning-algorithms) suggests to not use the dropdown box, but to reference the variable name in the text input field right beside the dropdown box. If I do so, everything seems to work fine. Although I have no proof the parameter value has effectly been in use in the learner.

  1. What ist the difference between the dropdown box and the text input field?
  2. Is there a way I can validate that my variables were in use in the learner? I'm already aware of loop stepping.
2

Answer:

1. The dropdown box selects the flow variable to assign to the node parameter value. The text field defines the name of a new output variable which will contain the parameter value used after execution of the node.

2. Use the text field to define a new output variable which will contain the value effectively used. Evaluate it e.g. using the "Variable to Table Row" or other nodes on the output variable port.

Btw. https://tech.knime.org/forum/knime-general/parameter-optimization-and-dialog-options-in-learning-algorithms contains a small typo: "nem" should possibly read "name"...