Complex task!

KNIME Analytics Platform
1

Hello all,

I have a table with 19 columns and 9 rows. I need to rank the rows for every possible combination (without repetition) of the columns. That is,

  • 19 combinations that have only 1 column
  • 171 combinations that have 2 columns
  • 969 combinations that have 3 columns
  • 3876 combinations that have 4 columns
  • etc.
  • 1 last combination that have 19 columns

The total combinations will be 524,287.

This process need to be repeated for another 9 tables. Across all 10 tables, differences in rows rankings will be calculated for every combination. The combination that yields the least differences wins.

I am desperately looking for a way to handle this complex task. Abundant thanks in advance. Hussam.

2

Hi Hussam,

I just played a little bitaround and you will definitely need to do some scripting to accomplish this tasks. What do you mean in differences in rows rankings? 

In case you are interested, this is the workflow I made. Than I realized the Rank node has a sanity check and hence my idea was no longer possible. 

Sorry for not being more helpful.

Best, Iris