Row-based filter, extract every X:th row

[Quaqq]

Hello,

I’m looking for a way to filter out every x:th row from a table, f. ex. to reduce a table by a factor.

I was thinking of using the row-based filter and an expression like
$$ROWINDEX$$ % 10 = 0 => TRUE (i.e. modulo of rowindex, this would keep every 10th row)
but I can not get it to work. Any suggestions on how to achieve this?

Thanks,

[aworker]

Hi @Quaqq & welcome to the KNIME community

This can be easily done using either the -Row Sampling- or the -Partitioning- nodes:

Row Sampling

Partitioning — NodePit

The configuration is the same for both and as follows to sample for instance 1 every 10 rows:

Hope it helps.

Best,

Ael

[gonhaddock]

Hello @Quaqq and welcome to KNIME forum
Alternatively to @aworker proposed solution you can use:

1. Math Formula node :
   mod($$ROWINDEX$$, 10) + 1

2. Rule-based Row Filter node (include TRUE matches) :
   $new column$ = 10 => TRUE
TRUE => FALSE

BR

[Quaqq]

Thank you @aworker and @gonhaddock for the solution(s), and also for the welcome.

Both solutions are appreciated, it’s good to know that there are dedicated nodes as well as a more generalized way to solve it. I have marked Ael’s answer as the solution, but both are valid. (As a sidenote, I guess it’s more efficient to use the dedicated node(s) when applicable, as there is no need to create an extra column and an extra step.)

Thanks again,
Q

[aworker]

Thank you @Quaqq for having validated the solution.

As mentioned in your last post of the thread, using a dedicated node is often the fastest solution, specially if the alternative to the dedicated node is a combination of nodes. The reason is that the same table needs to be scanned as many times as nodes are serially connected.

Having said this, one can monitor the efficiency of a node or set of nodes using the following monitoring node:

Timer Info

This may also allow you to compare @gonhaddock solutions to the -Row Sampling- and -Partitioning- nodes too.

Hope this helps too

Best
Ael