What inspired this Wish List Request? This is my personal suggestion, not a request from our users.
Please describe the new or improved feature or solution you would like to see.
My suggestion is "relative" profiler. The usual profiler draws prediction values. The relative profiler draws difference (or ratio) to a predicted value at a reference x-value.
For linear models (including generalized linear models), the prediction values for linear predictors are lp(x) = f(x)b, and the one for x0 is lp(x0)=f(x0)b. The relative profiler which I now suggest draws f(x)b - f(x0)b = (f(x) - f(x0))b. The standard errors is (f(x) - f(x0))`*covb*(f(x) - f(x0)), so confidence interval can also be calculated easily.
I think this feature may be useful for logistic regression (log odds ratios) and Cox proportional hazard model (hazard ratios) with using spline curves. See the following unsolved questions.
https://community.jmp.com/t5/Discussions/cubic-spline-with-survival-data/td-p/29270
https://community.jmp.com/t5/Discussions/Restricted-cubic-splines-with-survival-data/td-p/76290/page/2
I think that this idea itself is smilar to Multiple Choice Profiler on the Choice platform.
Note that this feature useless for many models. For example, relative profiler for main-effects-only models are just straight lines.
I also attach a JMP data table which shows an example for logistic regression with spline curve.
What would the value be to you and/or other users if this new feature is provided? The restricted spline curve for logistic models and Cox proportional hazard models is relatively popular in medical area.
Is this an external customer request?
No, this is just my personal suggestion.
Is there a competitive software package that has this feature or solution that we should be aware of?
At least, R and STATA support restricted spline curves for odds ratios and Cox proportional hazard models.
Will this impact a renewal, expansion, or adoption at a customer, and does this put an account at risk if we do not address this? No.
