Re: [Gretl-users] Variable selection
On 17.06.2018 09:03, Riccardo (Jack) Lucchetti wrote:
On a personal level, I agree 100%. As Clive remarked, the
statistical
issues implied by automatic model selection are indeed thorny. The
only convincing framework in which you can (to some extent) substitute
human judgement with CPU cycles is BMA, but apart from that I myself
would never base any critical statistical procedure on tools like
stepwise regression.
I near future (next week probably) I will upload the BACE
package. BACE
is just simplification of classical BMA (less restricted and ready to
use for non-stationary time-series) but gives some Bayesian look for
variable/model selection.
Marcin
--
Marcin Błażejowski