Adjustable Lag Difference?

[#1]

Don't know how to title this but here goes. I have data sets with several time points per batch. For each batch I want to calculate the difference in results from time point zero. I used to compare each result to column min or max (as suitable) which worked fine, but when results are a bit jumpy, the max or min may happen at another time point than zero.

I want the (desired) diff as below. Is there a simple way to do this in a formula? 

jmp 17.1

[#2 · Accepted solution]

If the results are ordered, you could use something like this:

:Result - :Result[Col Min(Row(), :Batch)]

Col Min(Row(), :Batch) will get the first row for specific batch and then we can access that row by using :Result[rownr].

If the data isn't always ordered it isn't really a problem, but it will require different formula

[#3]

Thanks for extremely fast, and useful, response! I will ponder how, but I see it works!

[#4]

Easiest way (in my opinion) to see how it works is to divide that formula into three different formulas

First one (BatchMinRow)

Col Min(Row(), :Batch)

Second using the just created BatchMinRow (BatchMinVal)

:Result[:BatchMinRow]

and finally third one

:Result - :BatchMinVal

this makes it much easier to see what the different parts in the formula do

[#5]

Just adding few possible options if the data isn't always sorted.

Using matrix calculations to get the indices

As Constant(
	times = :Time << get as matrix;
	batches = :Batch << get as matrix;
);

batch_rows = Loc(batches, :Batch);
batch_first_time_row = batch_rows[Loc Min(times[batch_rows])];
:Result - :Result[batch_first_time_row];

and using Col Sum to get the minimum value with comparison and product (if :Time isn't the smallest multiplier is 0).

:Result - Col Sum((:Time == Col Min(:Time, :Batch)) * :Result, :Batch);