Problem with new Random Forest Learner node (KNIME 3.2): variable number of attributes not allowed?

KNIME Analytics Platform
1

Hi KNIMErs

I recently updated to KNIME 3.2 and I am facing a problem with the new Random Forest Learner (Regression) node. I have some workflow where this node is inside a loop where the number and/or identity of the attributes from which the node should learn are changing from time to time. In the deprecated Random Forest Learner (Regression) node I was using the “Always include all columns” option of the attribute selection (indicating the response by a flow variable), and this was working well. With the new version of the node this option is no longer available. I try to make the node selecting all the input attributes in all the following ways:

  1. Manual selection: enforce exclusion of an empty list
  2. Wildcard/Regex selection: using patterns to select all attribures (e.g. “*” wildcard or “.+” regex)

In either way the node stop when the number/identity of the attributes change from one loop iteration to another provoking the stop of the whole workflow. Please, can you explain how to deal with this issue?

Thanks in advance,

Gio

2

Hi Gio,

let me quickly validate this with the developers. This seems to be the same problem as here https://tech.knime.org/forum/knime-users/knime-32-issuebug-in-tree-ensemble-learner-node-missing-option-include-all-columns

We will roll out some minor bug fixes (including this one) asap. i will let you know as soon as it is online.

Best, Iris

3

Hi Iris,

Thank you for your quick answer. I wasn't aware of the forum thread you mentioned... and yes, is the same problem. I would appreciate if you can update even this thread once the fixes will be available.

Thanks

4

Hi Gio,

sure i will let you know as soon as it is available.

Cheers, Iris

5

Hi Gio,

it was just published, you get it only via the update mechanism.

https://tech.knime.org/changelog-v32

Cheers, Iris

6

Hi Iris,

You are incredible guys, so quick!

I tested the update and now it perfectly works.

Thank  you