I am very new to PMML...
I have one dataset to be scored using two separate PMML files, one file can score part of data (age<=35) and another file can score another part of data (age>35). I know how to do it separately, but the question is :
How can I combine two files into one and score data?
Easiest method is likely to use a row splitter and different predictor nodes on each branch and then merge back together if needed with a concatenate node.
PMML actually supports model composition natively. (www.dmg.org/v4-1/MultipleModels.html) But I don't think Knime supports this yet, so if it absolutely has to be one PMML (for another program perhaps), you might have to convert to xml, fiddle around, and then convert back. Or use something external.