When manipulating data there are always things that are done per nominal value. This can be applied as demonstrated here: The workflow splits data based on a String column, manipulates the data separately, and then rejoins the data.