[Gnuplot-info] [Non linear regression] How to store the standard uncertainty of a parameter in a variable ?

[David K. <dav...@gm...>]

Hello,

I often use GNUplot to make regression using the "fit" function because
I find it useful and somehow easy to use. Since you define the
parameters as variables before to launch the fit, you can then print
out the values of the parameters you obtained after the fit.

My question is : how can you do the same with uncertainty errors ?

When you do a fit, if everything goes fine, you should end up with some
lines like :

 After 3 iterations the fit converged.
 final sum of squares of residuals : 7.15422
 rel. change during last iteration : -2.85539e-15

 degrees of freedom    (FIT_NDF)                        : 898
 rms of residuals      (FIT_STDFIT) = sqrt(WSSR/ndf)    : 0.0892571
 variance of residuals (reduced chisquare) = WSSR/ndf   : 0.00796684
 
 Final set of parameters            Asymptotic Standard Error
 =======================            ==========================
 
 BG              = 0.141662         +/- 0.002977     (2.101%)

My question would be, how to catch the number the asymptotic standard
error (that is to say the number 0.002977) in a var, to be able then to
print this number out.

Thank you very much.

David