What inspired this wish list request?
If my input (X) is non-linear in Fit Model, I have to manually try a variety of transforms (log, sqrt, recip, etc.) if a quadratic fit isn't appropriate or isn't the "best" fit.
What is the improvement you would like to see?
I want JMP to automatically try a range of X transforms and tell me which one(s) are best, similar to the BoxCox transform for Y. This could be in Standard Least Squares or GenReg within Fit Model.
Why is this idea important?
Help me improve my models and explore X transformations more efficiently without having to use a different platform like Predictive or Specialized Modeling which are often less interpretable.