I find the row splitter fairly slow and I need to split my dataset multiple times as my analysis is different per category in the data set. Now I’m creating like 15 row splitters splitters straight after each other.
It would be great if the row splitter could split the data in more than just 2 output tables. Basically similar like the concatenate node than can deal with multiple inputs, just with output tables.
What would make it even better would be if the splitter would also come in the form as the new expression row filter with an expression for each output table.
You may be interested in the following post (and the whole thread).
The result of the above was a couple of components which might not quite do what you want, but can go a step to reducing “row splitter chain hell”!
Both of these components work on the basis of the value of a “partition column”. So if you already have a column in your table that you are splitting by, you can use that. If not, you could create the partition column using a rule engine or other node and then follow it with one of the components:
Each produces a fixed set of outputs. Unfortunately there’s no option for dynamically adding outputs on components, but you can chain the last output to a further splitter if you need more than 10… still chaining but greatly reduced over the regular two!
The first one auto-partitions based on the column, using the values found. The second gives more control by allowing a regex pattern per output.
(I cannot vouch for performance though, as internally they also chain a series of row-splitters)
