Hello, Ingo, thanks for the feedback! At the moment I do not plan specific support with the module for Gnuplot's regression functions -- though you can invoke them yourself using Gnuplot commands if you like. The plot options "topcmds", "extracmds", and "bottomcmds" give you direct access to Gnuplot itself -- even to construct your own plot command line. I am not opposed to using this part of Gnuplot, I am merely busy with other projects too - so if you want to contribute a patch I would welcome it!
Kindly, Craig On Mar 14, 2013, at 8:51 AM, Ingo Schmid <[email protected]> wrote: > Hi, > > I've moved to P:G:G recently and enjoy it greatly. I used gnuplot for quite > some time for a) producing figures for publications and b) simple fitting > purposes. > > Now i see there is probably little point in accessing the fit function from > within PDL, although it is not quite that simple to do non-linear fitting PDL > with a function like > > f(x)=g0*(x-x0)**g1*exp(-g2*(x-x0)+s0/(1+exp(-s1(x-x0-x1))* > > Doing it in gnuplot it is very convenient, though. Is this planned in a > future release? > > The more important point a), though is, when i.e. doing linear regression of > some simple x-y plots, I miss the simple > > gplot ('$a*x+$b'); > > functionality. You have to invoke xvals and send it to gnuplot, which appears > awkward. Or am I missing something? > > Here is some real world example, $z is $y->fill_m to remove missing values > (expressed by BAD). > > %m=$z(,8;-)->ols($z(,27;-)); > gplot (xlabel=>'pH_{end exercise}', ylabel=>'TTP [s]',term=>'postscript enh > 16', out=>'pH_ee-ttp_bold.eps', > with=>'points ps 2 pt > 5',xrange=>[6,7.5],yrange=>[0,500],$y(,27;-),$y(,8;-),with =>'lines lw 1' > ,y=>$m{b}->(1)+$m{b}->(0)*xvals(8), {lw=>.1}); > > > > Regards > > Ingo > > *In case someone is interested, it's a gamma variate plus a signoidal (from > one of our publications > http://onlinelibrary.wiley.com/doi/10.1002/jmri.23919/abstract). I have used > PDL::FIt::Levmar and done the PP function calls, I guess P::F::LM can do so > as well. > _______________________________________________ > Perldl mailing list > [email protected] > http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
_______________________________________________ Perldl mailing list [email protected] http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
