How to compute mathematical functions based on condition of the column names

[gonhaddock]

Hello @RoyBatty279 and welcome to the KNIME forum.

@RoyBatty279

As you can see, each column represents one variable at a given timepoint (in raw values).
This is an oversimplification, so in reality the workflow should work for many variables and many timepoints.

Based on your comment, what you show us is a nested hierarchy in your column header name. You need to transform it in the way that time-series ($Day$) arrange in a column, then: $Variable_num$ (character), $Baseline$ (value) and $Raw$ (value).

As all your values are numeric you won’t find many hurdles to do it. Maybe unpivot separately Baseline and Raw; then work the string $Column_Names$ in the two separated tables and a final Joiner with double matching criterion ($Day$, $Vasriable_num$)

You can check this post for insights:

Convert multiple columns for each year of data into a single column and date column

Hi @zm004433 and welcome to the forum. Try this workflow out and see what you think. It works on your sample data, but I think it should be fairly extensible to your larger dataset. The trick is making sure you understand how the configuration of the Unpivoting and Pivoting nodes work. [2021-06-16 14_37_03-KNIME Analytics Platform] [2021-06-16 14_35_31-Pivot table - 0_6 - Pivoting] UniversityDataUnpivotExample.knwf (12.3 KB)

The final Math Formula will be just the difference of two columns. You can later rearrange back the data as your preference.

BR