Rsquare disappear in Nonlinear Fit model

[#1]

Hello everyone,

I have a problem with the Nonlinear Fit platform.

I am testing a broken line model on simulated data.

The model converged and I get a solution with Rsquare and estimates of the parameters.

I needed to have the confidence limits so I clicked on the confidence limits button.

The confidence limits appeared in the parameters table but the Rsquare disappeared !

I could have the Rsquare back when I launched again the computations with the Go button but then the confidence limits were gone.

So I can have a nice report but I have to choose between Rsquare and CL. Of course I can export the results as tables but it is a bit frustrating though.

Is there a solution to this?

[#2]

I will start with I don't actually know the reason, but I have some thoughts...and I could be completely wrong

R-square is the coefficient of determination and assesses how much of the variation in the data set is explained by the model.  With just the model effects, the R-square is reported, however, when you take into account the variation that is unassigned to the model with confidence intervals, essentially all of the variation is explained so the R-square is no longer needed. (or is 1).

[#3]

It looks like there was a bug here. Your post got sent to the development team, and i can confirm the behavior has been addressed in JMP 17. (JMP 17 will be released in the fall of 2022.) Thanks for reporting your issue here; these forums are watched and issues like this are often reported to the development team and fixed.

[#4]

Audrey, please see my post above (or below) that I just posted.

Thanks

[#5]

I too have a similar problem with JMP 18 using the nonlinear platform with an equation similar to the Rodbard 4PL equation, where if I choose to "Lock" one of the parameters (the lower asymptote in my case), then the RSquare disappears.

  

Can this also please be addressed?