Subgroup size in Nonlinear Modeling

Question

Hi all,

I am using Analyze/Specialize Modeling/Nonlinear with big class.jmp

If I make a linear model, Alert window displayed.

My questions are

1. Among of six subgroups, the size of five groups is at least 2.

Is there any method that can execute modeling for 5 subgroups except subgroups that subgroup size 0 or 1 ?

I mean some method that can execute modeling from the beginning without extracting corresponding row(data).

2. Can I make model  only for subgroup that data size is equal to or greater than a certain number ?

Thanks

IkjuShin · Answer

I excluded small-size subgroups using Pareto before using nonlinear analysis.

IkjuShin · Answer

Hi Dan

Thank you

Graph builder is good method in this case and I also used by(sex), local data filter(age) in Fit Y by X

In nonlinear modeling, if we select like below, we could not make model

because there is one subgroup that data size is just one. that is my concern.

Dan_Obermiller · Answer

How are you subgrouping? When I put the Age as a By variable, I got models for every group in Fit Y by X as well as in Nonlinear. In Fit Y by X the Ctrl broadcast command worked and a line was fit for each age group, even the one with only one point (it was just a mean model). In the Nonlinear Platform I had to hit "Go" for each group, so I could have easily skipped the single point group, if desired. But even there I did get a nonlinear model (it only adjusted the intercept and left the initial value for the slope).

You could even get something from Graph Builder using a linear model:

Would this approach possibly work for you?