Bert,
Thank you for your thoughts.
I can assure you I have plotted the data back and forth many times, and
that overfitting has nothing to do with it. This is not a _statistical_
problem, but a _technical_ problem. Something that works well in ANY
reliable statistical software doesn't work
?nls.control
fit- nls(MFI~a + b/((1+(nom/c)^d)^f), data=x, weights=x$weights,
+ start=c(a=100, b=1, c=100, d=-1, f=1),
control=nls.control(warnOnly=TRUE))
Warning message:
In nls(MFI ~ a + b/((1 + (nom/c)^d)^f), data = x, weights = x$weights, :
step factor 0.000488281 reduced below
Thanks, Keith.
I failed to cc the following reply to John Nash to the list. Your
email persuaded me that it might be useful to do so.
None of this changes the fact that the model is overfitted. You may be
able to get convergence to some set of parameter estates, but they
won't have much meaning
On Wed, May 2, 2012 at 3:32 PM, Michal Figurski
figur...@mail.med.upenn.edu wrote:
Dear R-Helpers,
I'm working with immunoassay data and 5PL logistic model. I wanted to
experiment with different forms of weighting and parameter selection, which
is not possible in instrument software, so I
Dear R-Helpers,
I'm working with immunoassay data and 5PL logistic model. I wanted to
experiment with different forms of weighting and parameter selection,
which is not possible in instrument software, so I turned to R.
I am using R 2.14.2 under Win7 64bit, and the 'nls' library to fit the
Plot the data. You're clearly overfitting.
(If you don't know what this means or why it causes the problems you
see, try a statistical help list or consult your local statistician).
-- Bert
On Wed, May 2, 2012 at 12:32 PM, Michal Figurski
figur...@mail.med.upenn.edu wrote:
Dear R-Helpers,
6 matches
Mail list logo