[SOLVED] ROWNUMBER() in RapidMiner

Mario_Hofmann

Hello,

currently I m trying to get my whole ETL for a process in Rapidminer. I ve got a table (A & like:

colA | colB | colC
A    | 2      | 1
A    | 4      | 2
B    | 1      | 1
C    | 2      | 1
C    | 3      | 2

Is there a way to create column C in rapidminer? In MS SQL I can use Row_Number and Partition (http://msdn.microsoft.com/de-de/library/ms186734.aspx) . I found a way using Loop Values, but this is far too slow for a table with several 10ks of examples.

Regards,

Mario

MariusHelf

Hi Mario,

what are you doing in Loop Values? Actually, to create an index for each unique value in column A, there is no other choice currently in RapidMiner than using Loop Values. Inside, you could use Generate ID, which should be quite fast.

Best regards,
Marius

Mario_Hofmann

Hi Marius,

I was actually already doing the filtering inside the loop, just to generate the id is of course a much better idea. But I am close to finalize my process, I will report on the results.

Regards,

Mario

MariusHelf

Well, of course you also have to filter before generating the id, otherwise you get an id for the complete dataset, not only for the current value oO

Mario_Hofmann

Sure thing, there are actually some things I remember from your lessons.

dsh_shadi

Thank you Marius Helf,  your reply was so useful