R help - Mar 2017 - nlmrt problems - No confInt, NA StdErr, t-, or p-values

Dear list,

I want to use nlxb (package nlmrt) to fit different datasets to a gaussian,
obtain parameters (including standard error, t-and p-value) and confidence
intervals.

nlxb generates the parameters, but very often results in NA standard error,t-and
p-values. Furthermore, using confint() to obtain the confidence intervals
generates a : Error in vcov.default(object) :   object does not have
variance-covariance matrix" erro.

Can someone indicate why is nlxb generating NAs (when nls has no problem with
them) and how to obtain confidence intervals from an nlmrt object?

Thanks


	[[alternative HTML version deleted]]

You can use wrapnls from the nlmrt package to get an nls object.  Run
it instead of nlxb. It runs nlxb followed by nls so that the output is
an nls object..  Then you can use all of nls' methods.

On occiasion that fails even if nlxb succeeds since the nls
optimization can fail independently of nlxb.  Also, it does not show
the output from nlxb, only from the final nls, so you could
alternately run nlxb and then run nls2 from the nls2 package after
that.  nls2 can compute the nls object at a particular set of
coefficients so no second optimization that could fail is done. Here
is an example that uses nls2 to generate starting values for nlxb,
then runs nlxb and then uses nls2 again to get an nls object so that
it canthen  use nls methods (in this case fitted) on it.

http://stackoverflow.com/questions/42511278/nls-curve-fit-singular-matrix-error/42513058#42513058




On Tue, Mar 21, 2017 at 6:57 AM, DANIEL PRECIADO <danprec at hotmail.com>
wrote:
> Dear list,
>
> I want to use nlxb (package nlmrt) to fit different datasets to a gaussian,
obtain parameters (including standard error, t-and p-value) and confidence
intervals.
>
> nlxb generates the parameters, but very often results in NA standard
error,t-and p-values. Furthermore, using confint() to obtain the confidence
intervals generates a : Error in vcov.default(object) :   object does not have
variance-covariance matrix" erro.
>
> Can someone indicate why is nlxb generating NAs (when nls has no problem
with them) and how to obtain confidence intervals from an nlmrt object?
>
> Thanks
>
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.


-- 
Statistics & Software Consulting
GKX Group, GKX Associates Inc.
tel: 1-877-GKX-GROUP
email: ggrothendieck at gmail.com

Note that the recently released package nlsr by Duncan Murdoch and I has a
slight update in the functions nlxb() and
nlfb() as well as a lot of features for symbolic derivative calculation.

JN

On 2017-03-21 06:57 AM, DANIEL PRECIADO wrote:
> Dear list,
>
> I want to use nlxb (package nlmrt) to fit different datasets to a gaussian,
obtain parameters (including standard error, t-and p-value) and confidence
intervals.
>
> nlxb generates the parameters, but very often results in NA standard
error,t-and p-values. Furthermore, using confint() to obtain the confidence
intervals generates a : Error in vcov.default(object) :   object does not have
variance-covariance matrix" erro.
>
> Can someone indicate why is nlxb generating NAs (when nls has no problem
with them) and how to obtain confidence intervals from an nlmrt object?
>
> Thanks
>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>