cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
JMP is taking Discovery online, April 16 and 18. Register today and join us for interactive sessions featuring popular presentation topics, networking, and discussions with the experts.
Choose Language Hide Translation Bar
l_yampolsky
Level III

construct a loss function for nonlinear fit

I see a similar question with no answer, so not much hope but let's try.

Let's say I want to do a non-linear fit (for example logistic function) and then use either log-ratio test or AIC to demonstrate that it fits better than linear. I understand that I need to construct an appropriate Loss function and then the Solution report will show the log-likelihood. What do I put into the loss function?

Is there another way to obtain LL of a non-linear fit?

Thanks!

3 REPLIES 3

Re: construct a loss function for nonlinear fit

I recommend you see Help > JMP Documentary Library > Predictive and Specialized Modeling > Chapter: Nonlinear Regression > Section: Example of Maximum Likelihood: Logistic Regression. The procedure and process are fully explained and an example is given that results in -2L for the loss value of the fitted model.

l_yampolsky
Level III

Re: construct a loss function for nonlinear fit

yes, I have seen that example. However, it provides instructions of how to test the hypothesis about logistic regression (i.e., with categorical response variable). I am trying to compare two models with continuous response variables - one linear, one non-linear.

But thanks!

l_yampolsky
Level III

Re: construct a loss function for nonlinear fit

my problem was solved by moving from JMP10 to JMP15 (for now just as a trial). There there is a Curve Fit dialog which does exactly what I need.