The formula editor is very important and helpful for my work in general.
I would appreciate more ease and flexibility for building formulas across several rows not just within the same row.
For repeated measures of variable length I would appreciate a function that collects the last value measured for each analysis unit (e.g.patient); this should be possible both in a horizontal and in a vertical arrangement of repeated data.
For complex formulas I would appreciate the possibility to store an interim result and then use this result in the final formula.
I often have a frustration where I've got a fairly complex formula almost complete, then in editing the JSL of it I make a mistake, maybe a parenthesis is missing, and it just blots out the whole thing and I'm back to square 1 instead of getting a chance to edit the JSL I had there.
At this moment, you can select whether to use Table columns, or table variables, constants, parameters, … through the drop-down menu at the very left corner of the Formula window. I guess the most widely used, is “Table Columns”. For me it would be very useful to have the other options (such as table variables, constants, parameters, ..) in a separate drop down menu, visually next to that of the Table Columns, so you can easily build a formula using table columns and table variables, without having to click on the drop down menu each time you switch from using a table column and a table variable.
And then a more general suggestion (which possibly goes beyond the formula editor): enable importing a list of table variables, eg data table with two columns (name of variable & value of variable); these variables are then added as table variables to your table. In this way you would have a clear overview of the variables/parameters you are working with, and can easily use these values for your calculations. Additionally, they should automatically update when you change the values in the initial table. Alternatively, the formula editor could allow using values from an external data table which lists all the variables (and then obviously, when changing the values in this external table, the formula editor also uses these new values).
Add the PERT function to the formula editor (now you have to manually calculate the alpha and beta values and then use the BETA distribution)
One other thing that would be nice, when scripting in formulas through JSL, if I use a variable that was in my JSL code, it often simply gets sent to the formula as that variable. In the pasted view below, beta and eta are values determined in the JSL, then fed back to a column formula, which appears as,
What would be helpful is for these to either expand to their actual numbers, or have them appear as something that is selectable, or at least something that shows what the values inside the formula are at that moment. I have looked in the drop down that lists "parameters", but they do not appear there, so oftentimes, once the formula is written, it's tough to tell what the variable values are.
One thing that I find would be useful would be a way to select multiple columns with the ability to select a function that will act on all of them.
i.e. say you have a wide dataset with 100s of columns and you want to get the sum of all of them without stacking the dataset.
Unless there is a way that I am not aware of, you need to insert each column one at a time separating them by clicking '+'. This would be a feature request that I would like, unless there is already a solution.
jason.paquette, indeed there is already a way to do what you ask.
Add the function Sum() to your formula. Click on the first column that you want to include and then shift-click on the last one. JMP will add all the columns in between.
Most of the functions in the Statistical list support this.
Another option, new in JMP 12, is to select the columns of interest in the data table. Then right-click the header of one of them and select New Formula Column->Combine->Sum. This provides the option to select columns that aren't next to each other in the table by ctrl-clicking.
I've been thinking about this one for a while.