with the latest KNIME releases (4.1.1 and 4.1.2) I got some problem with the ‘Coluimn Filter’ and ‘Column Splitter’ nodes when trying to define the exclude/include lists via Flowvariables as the field array/list below the e.g. ‘excluded_names’ node (in the Flowvariable menu) does not appear anymore!
The funny thing is: In WFs I created with older versions this list still appears when executing under 4.1.2 - but when I take e.g. a new Splitter Column node and try to go for the same settings this is list not shown/presented anymore.
So obviously the functionality is still available, but is NOT presented in the UI/menu anymore.
The workaround with ‘Reference Column Filter/Splitter’ nodes is somehow cumbersome - so hopefully you can provide a fix soon.
On Windows I don’t have this issue. While waiting for a KNIME team member to check this issue, I have a tip for you which may solve this.
Sometimes after a few hours of working with KNIME, I have experienced some issues with configuration windows in displaying options. For me resizing the configuration window works, maybe it does for you as well.
Yes, I can see this options - but nothing more and this is my problem!!
To define the e.g. ‘excluded_names’ Columns, in earlier versions you got one String Editline per column in the excluded area - you had to definein advance in the ‘Column Filter’ dialog. see jpg attached!
thx for clarification:)
Unfortunately, these new Flowvariable types (e.g. arrays) introduced with 4.1 are not available in the ‘Java Edit Variable’, ‘Java Snippet’. etc. nodes.
My Challange: I’ve created a couple of nodes (components) for AutoML where I have to know (amongst others) the name of the class-column (the attribute storing the lable to be learned). As I’m applying these nodes on different data sets, I used a ‘Java EDit Variable’ node for all of these settings. So I created a variable ‘nameOfClassColumn’ containing the name of the specific attribute.
My Problem: The improvements introduced with 4.1 obviously caused the ‘deactivation’ of the older way-of-working without providing sufficient support - at least for my situation. Now I’m forced to use a couple of nodes to get a simple String-Array as a Flowvariable.
Sample: If I want to filter out (exclude) the column of a table which’s name is stored in the variable ‘classColumnName’ I have to:
1.Create a table with one String array column (e.g. Java Snippet)
2. Use the Table-Row-To Variable to Convert the array-column into a array flowvariable
3. Use this new Flow variable in the Column Filter node
Knowing that in some cases step 1 requires more than 1 node this process is somehow cumbersome.
However, I would appreciate very much that the ‘Java Snippet’, ‘Javav Edit …’, etc nodes are extended so that the new Flowvariable types can be used too!