KNIME missing value node does not work

Question

I'm running KNIME to prepare some ML data sets. It is about ecommmerce transactions where a association rule analysis should be done later. But my current problem is rather simple, I am in a 'data preparation step' where the missing value node in KNIME does not work.

I'm running KNIME 3.5.2 on mac osx

The sample workflow is primitive and looks like:

config of the first node which reads the excel file (XLSX)

Config of the second node , which should replace the missing values:

But the annoying result is:

The warning being shown is:

WARN Missing Value 2:2 The current settings use missing >value handling methods that cannot be represented in PMML 4.2

The warning being shown is just about the possible problem with reusing the KNIME node in a machine learning context as with R oder Spark/MLLIB because if model incompatibility problems. This can be ignored here.

I have to say, yesterday this has worked. I changed basically the XMX setting in KNIME.ini, restarted KNIME and have run several other workflows. When I came back it did not work anymore.

I think KNIME has some kind of state in this node, at least obviously in the eclipse workspace for sure. I have started KNIME already mit -clean, nothing helps. Running this sample on the same KNIME version on a colleagues computer works fine.

Any ideas ? thx (btw: I know that I could easily do the same job with a few lines in R oder spark...)

Answer 1

In KNIME it's only a proper Missing Value if you see a "?" in the Knime table.

Answer 2

"Solution": I have changed from a Excel File Reader to a CSV Reader. The Excel file obviously contained some kind of "value" in the cells which are shown as empty.