I tried to find a similar problem in the forum but couldn't find anything.
I'm using the below workflow, where I apply the "M5 algorithm to my data".
My goal is to do some sort of "forward selection". I try to apply this algorithm first to only the first 11 rows of my table, then to twelve rows, then to 13 etc in a loop. Each time I compute the error. (The goal is to find out how many observations are necessary)
I am not able to configure the "row splitter node" so that it selects in the first loop 11, then 12, then 13 rows etc.
Thanks a lot!
You can use the currentIteration flow variable from the loop node for that. It should already be present in the Row Splitter, so all you have to do is to chose include rows by number and to set RowRangeEnd in the Flow Variables tab to use currentIteration. If you want to start with more than one row in the first iteration or use a different step size than 1, you have to compute your own variable based on currentIteration, either via a Java Edit Variable or by converting it to a table, computing, than converting it back.
Hi, I would have done something like below in attachment, controlling the Flow variables RowRangeEnd with the current iteration.
Thanks a lot , these answers were very helpful!