Group by - sooo useful!!! how about Interpolate?

[hogi]

☑ cool new feature
☑ could help many users!

☐ removes a „bug“

☐ nice to have

☐ nobody needs it

What inspired this wish list request? 

The Group By concept in Jmp is sooo useful!

You can find it everywhere. e.g. in Graph builder one can use X/Y groups to generate split graphs - and summary statistics will automatically give the correct results for every single graph.

In Platform menus, there is a By option - which splits the data into groups and every group is analyzed separately.

Under the porcellain layer, there are more group by options, like the Col aggregation functions (e.g. Col Median(:height, :sex)) which automatically return separate aggregated values for groups of rows.

What is the improvement you would like to see? 

Please use such Group By options wherever it is possible.

let's see if other users find more places where Group By could be useful ...

here I want to focus on a Col Grouped Interpolate function which can be used in a Column Formula like

New Column("interpolated", Formula(Col Interpolate(:newX, :height, :weight, by(:sex))))

It should generate an interpolation for :weight by :height - split by subgroups M/F in :sex - and use the interpolation to calculate values @ x values :newX.

In a similar way, the existing Interpolate function could be lifted to a new level by providing an argument  BY() which can be used to interpolate subgroups
[for this case, I think an additional argument for()is needed to specify the groups of interest and their order]

Grouped Interpolate 

Interpolate(myheight, :height, :weight, by(:sex), for({"F","M"}))

Definitely worth to have it - which can be seen by the fact that users invested the time to program their own grouped Interpolate to get the results:
https://community.jmp.com/t5/Discussions/Grouped-Interpolate/m-p/692358/highlight/true#M87807 

Why is this idea important? 

Group By is one of the concepts which makes Jmp a million times more useful for exploratory data analysis  than Excel .

use this concept at as many locations in the program as possible.

more wishes submitted by  

[hogi]

Besides a siple interpolation, one could also think about more fancy fits to the data - along the idea of saving a Spline formula to a data table - such that the calculated values can be used for further processing.

The disadvantage of saving the spline [like it's already possible at the moment]:
it's static !

If values in the table are changed or added, the formula will not update.

[hogi]

other functions which will benefit from ByVar:
Dif() and Lag() formula functions by column group (byVar argument)