Hi, I am having some problems with gigFit and would like confirmation on
other platforms; mine is mint; basically Debian.
Although I got a good fit for the density function with the GIG equation
in another curve fitting program, I would really like to use R's tools
for confidence intervals and manipulations; but the problems below make
me uneasy.
Problem one (from examples with parameter change):
Code:
paramStart.X2002<-c(2.044, .622, 2)
param<-paramStart.X2002
dataVector <- rgig(500, param = param)
gigFit(dataVector)
Result
Data: dataVector
Parameter estimates:
chi psi lambda
0.007543 0.722962 2.535284
Likelihood: -644.6349
Method: Nelder-Mead
Convergence code: 0
Iterations: 104
Notice that the chi parameter is significantly off.
-----------------------------------------------
Problem two
I have data that I have fit to the GIG equation in a curve fitting
program with good fits and resulting parameters very close to the above
paramStart.X2002
The gigFit program on the same data goes completely off into strange
territory. I hope that fixing the above problem would fix this.
----------------------------------------
Problem three:
When I do the above with plot
gigFit(dataVector,plot=TRUE)
I get a warning
Warning message:
"
In incompleteBesselK(x[i], y[i], -lambda, tol = ibfTol, nmax = nmax) :
Maximum order exceeded
Result may be unreliable
"
Ray
-- This space is for my comments on the world
(Intentionally left blank.)
______________________________________________
R-help@r-project.org mailing list
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.
