Re: [Gretl-users] Machine Learning in Gretl
Am 20.07.2017 um 14:22 schrieb Periklis Gogas:
Thank you Artur,
That would be very interesting. I am not familiar with that but with
the libsvm.
Interesting -- could you explain how familiar you are with libsvm? In
principle this looks like a good candidate to wrap, technically
speaking, because AFAICS it's a C/C++ library.
Apart from that in general I agree with Artur that wrapping some
existing approach in R might also be a good idea. (Assuming that in the
background the respective R package is also calling compiled C/Fortran
code.) I think it is utopian to think that we could or should re-invent
or port everything developed for R.
Obviously for the last couple of years R has had a huge momentum behind
it. I think this is good, even though in some sense we are competition.
cheers,
sven