Hi Matthieu,
I'm working on an out-of-sample, rolling window & adaptive fit predictive LSTAR model which uses weekly index logged returns. Initially I almost got it working with simply using the "lstar" function and some more code for the selection of the parameters in each window.
Now, since discovering the selectLSTAR function (bit of a backwards way I admit) for the automatic hyperparameter selection, I decided to test how it works. With the example from the package information it works well:
llynx <- log10(lynx)
selectLSTAR(llynx, m=2)
but using my own data I encounter the following error:
Error in seq.default(start.con$gammaInt[1], start.con$gammaInt[2], length.out = start.con$nGamma) :
'to' must be a finite number
Despite my best efforts, I have not been able to solve this issue on my own by seeking help elsewhere. If needed, I would be happy to provide a reproducible example for further inspection.
Thank you in advance!
-Alex G