topic Re: Extracting Confidence Interval Parameter Estimates Equations from a Generalized Linear Model in Discussions

Extracting Confidence Interval Parameter Estimates Equations from a Generalized Linear Model

Michael_Foster — Sat, 10 Jun 2023 23:48:49 GMT

Hello,

I'm trying to save the equation for the Lower CL and Upper CL (shown below) but I'm unable to.

As shown below, I am able to save the original model prediction formula and I'm also able to save the values of the mean confidence intervals, but even if you right click the column, there is no formula that is translated over with the confidence intervals. I'm trying to gather the formulas in order to input other numbers into it for prediction capability. I also tried copy pasting the numbers in the table above into the original probability formula but that did not work. For example, I took the intercept of -7.637 and substituted in -9.666 instead to reformulate the column, but the numbers didn't match even after I did that for every variable down the column. Any reason why that wouldn't work? Any help on extracting the confidence interval equations for the generalized linear model would be greatly appreciated. @Mark_Bailey JMP 15.2

Original model equation shown below that I attempted to substitute CL values in going down the column 1 for 1. All continuous except for 1 categorical factor.

Thank you,

Mike

Re: Extracting Confidence Interval Parameter Estimates Equations from a Generalized Linear Model

Mark_Bailey — Tue, 24 May 2022 18:39:08 GMT

You might add a request for this feature in the Wish List area of the Community. (Check first to see if it has already been requested and, if so, vote (kudo) for it).

For now, though, you can use GenReg for model selection, and then go back to Fit Model dialog and set up the same linear predictor. The Fit Least Squares platform is the result of using the default Standard Least Squares personality. This platform can save the formula that you want. See Save Column commands in the platform menu.

Re: Extracting Confidence Interval Parameter Estimates Equations from a Generalized Linear Model

Michael_Foster — Tue, 24 May 2022 19:08:40 GMT

Thank you very much for the reply Mark. I've taken a couple of your classes which is why I ended up tagging you. I'm trying to predict whether something is a 0 or a 1, so a nominal response, which gives me a limited number of options for "Personality" under the "Fit Model" function. I lose track of following along with your reply after you say to go back to the Fit Model dialog and set up the same linear predictor. Am I extracting information from the Generalized Regression model and then using that as an input for a different model type? I am unable to use Standard Least Squares (grayed out) since my response is binary (I'm assuming). Any other guidance given my situation?

Re: Extracting Confidence Interval Parameter Estimates Equations from a Generalized Linear Model

Mark_Bailey — Tue, 24 May 2022 21:07:54 GMT

The binary response changes everything. JMP can compute the CI, but it does not currently save them as a value or a formula.

The Prediction Profiler used within the fitting platform provides CI of the response prediction, if that helps.

Re: Extracting Confidence Interval Parameter Estimates Equations from a Generalized Linear Model

Michael_Foster — Tue, 24 May 2022 21:22:28 GMT

Thanks for the follow up. I'm happy to hear that I'm not crazy and that it can't currently be done within JMP. Closing out the thread, appreciate it!