I have some time course data of cell growth (mammalian) and would like to fit them to obtain cell division and death rate. JMP has a built-in model named "Cell Growth 4P" in the "Fit Curve" option which seems to work pretty well for most of my data. However, for one of the groups, the model showed negative division and death rates. The fit looks pretty good otherwise. I have a few questions on this:
1. Is there a way to set parameter bounds with the Fit Curve feature? I'd like to set the lower bounds for both of the above rates to be zero to avoid running into this same issue.
2. I also tried the Nonlinear Fit option since it allows me to specify parameter limits. However, I tried inputting the same equation as Cell Growth 4P above but could not get the fit to be anywhere as good as Fit Curve results. Most of the time, the fit parameters seems to move very little from the initial input values. What settings may I change to help with this?
3. On a less related note, is there anyway I can find the literature source for this Cell Growth 4P equation?
Regarding the nonlinear platform - did the algorithm converge on a solution? You could try increasing the maximum number of iterations, or relaxing the convergence criteria; alternatively using different start points based on domain knowledge. There is also an option to perform a grid-search to investigate alternative start points.
I'm attaching the before and after of an attempt here. I also included how the fit would look like using Fit Curve option.
The strange thing is, even if I put in the values derived from Fit Curve results, Nonlinear Fit would still converge to a wildly inaccurate fit. Increasing number of iterations doesn't do anything since it always seems to converge at the second iteration. I also tried reducing the Convergence Criterion but that doesn't help either.
Is it possible for you to send the data exhibiting the problematic fit and the non-linear model that you use to JMP Technical Support (firstname.lastname@example.org)? I asked them about the behavior you encountered but they don't see it with their examples.
There are no labels assigned to this post.