Fit Y by X - Multi parameters - Saving Individual Probability Formulaes

[#1]

Hello ... 

I am performing logistic regression via Fit Y by X platform where I have hundreds of parameters. 

How do I extract probability formulae from all fits ? I tried w/ natural instincts of [ctrl - save probability formula] and I think it may cycled through the the parameters, however at the end there was only a single formula created (even it created prior ones, they must have been overwritten). 

Feeling that this may yet be another of those things only possible via scripting, could you give me a simple script start for something like this - unless of course if there is a practical way of doing this fast for all hundreds of parameters at once. 

Using JMP Pro 17

Appreciate the support. thanks

[#2]

I did not find any overwriting of the data.

Using the JMP Sample data table, called Probe.jmp I ran a Fit Y by X, specifying DieX as the Y, Response and 37 numeric columns as the X, Factors

Holding the CNTL key down, I clicked on the red triangle and selected   Save Probability Formula

The data table started out with 394 columns, and ended up with 616.

Each of the 37 parameters created 6 columns,. Each set of columns are unique and based upon a different Y response for each of them.

What did I do that was differant from what you did?

[#3]

hi Jim, 

What I am doing is different on two fronts: 

a- mine is multinomial (have 5 classes) and likely yours is binary - what that should not matter 

b- mine is By a group

I tested w/ 2 parameters and no "By" group ... it worked as you did show

I then tested w/ 2 parameters and "By" group .. it did not work. In the end result, there was only one set of probability that incorporated "By" correctly into the formula (instead of creating one formula for each of the "By" options) but it failed because it skipped first parameter and only did this for the second one. 

[#4 · Accepted solution]

I replicated your findings.  It looks like you will have to either subset your data by the By variable, or by the different parameters to get your results.

[#5]

Thanks Jim.. To me this looks like a bug in JMP17 Pro. If you agree, would it be possible for you to convey it to the dev team ?

My current work around is to save my original file into 3 options (By group luckily does not have too many options), perform logistic fits w/o By, store all probabilities and then combine all results back together. 

[#6]

You have the ability to inform the JMP development team yourself.  Just go to the JMP WishList box at the top of the Discussion Webpage and click on it and enter your issue and how you think it should be handled.