Feature Elimination MetaNode


I am having some problems with this metanode, It has two input ports, with the first one I have no problem, but with the second, does it supossed to be the same input than the first one?

It needs a table with the same features that the used in the first input node, so if I connect the same node (with my data) to the two both input nodes, it should work I think… If i do this, I obtain a warn " Input table does not contain any of the columns used in the feature elimination", when I am using the same node, so i do not understand…

Any help please?

Thanks in advance,


Hi snyder,

The idea is that the first input port is connected to training data (with class information) while the second is connected to unlabeled data from which certain columns are filtered out afterwards.
The error message on the filter node is indeed misleading (if not wrong). It will go away, however, if you execute the feature elimination loop inside the meta node and select some features in the filter’s dialog. I will fix this annoying behavior.