Normalizer problem excluding columns

Hi! I have a csv file that has 12 numerical columns. At first six of them were integers, six floats. Normalizer approved only the integer columns. I changed the floats to integers, but still Normalizer does not approve them. It keeps the original column set even if I delete the Normalizer and drag it again.

source file now
1.;Surname1, Firstname1;HIFK;VP;1;3;5;-2;38;33;3;3;50;33;100;133

WARN Normalizer 0:31 Auto-configure: [0, 1] normalization on all numeric columns: “O”, “L+”, “L-”, … <3 more>

These are accepted:

O (Number (integer))
L+ (Number (integer))
L- (Number (integer))
Lpm (Number (integer))
HPA (Number (integer))
PPA (Number (integer))

How can I reset the Normalizer to read all columns?
Also, should Normalizer accept also float columns?


Hi @lassease -

I tried to recreate your problem, but so far haven’t been able to. A few questions:

  1. What version of KNIME are you using, and on which OS?
  2. How are you reading in your data? When I read your sample using the CSV reader, it detects all the numeric fields as integers.

But even if I change the numeric fields to double or long types, the normalization node still handles things properly. If you could upload a sample workflow, that would help.


Hi ScotttF

My KNIME is 4.3.2 and Win10.
I use a CSV Reader that to my eyes reads all fields. So just a CSV Reader and Normalizer.

In the zip file I have the example with two csv files. Pelaajat.csv is the original one that shows how the Normalizer can’t figure out all columns.
If you change in the CSV Reader to pelaajat3.csv (has only integers) you can see the behaviour of the Normalizer.

(the files are not uploaded because of the restrictions of this system: csv and zip are not allowed. I could email them.)


Thanks - please email them to me and I will take a closer look.

This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.