[Gretl-users] Re: BVAR pkg

Am 15.08.2023 um 14:59 schrieb Brian Revell: Well, Luca also has another package called "BayTool", and you may be interested in the included function BayLinMod(). Admittedly the GUI currently still is a little rudimentary, but maybe you could give it a try and report back. Apart from that, at the risk of sounding repetitive or annoying, the question is why you want to use Bayesian regression. My impression is that 90% of the applications just want to have shrinkage and regularization that Bayesian stuff gives you "for free". The frequentist alternative these days would be Lasso and Elastic Net, which gretl offers natively with its "regls" functionality, and with a super-nice GUI. Of course, if you truly have an idea about the prior distribution, or if you want to use it for teaching or for comparison, then a Bayesian regression method is useful, and your request is perfectly legitimate. thanks sven