Installation and Bayesian Econometrics support

[Filippo Pellegrino]

Hi all, 

I've been very impressed with Sage and I'm downloading the .dmg for my mac at this moment. 

I'm writing here mainly for two reason: 

- firstly because I would like to understand if and how I can set my current python directory as default; 

- secondly because I'd like to know If Sage or some of the python libraries installed have a direct support for bayesian econometrics. I'm thinking to use Sage to the writing of my master's thesis and I need to perform estimates with Bayesian VARs using noninformative and informative priors. If there is not a package for bayesian econometrics I would do one myself and I would happy to post it here if you need it.

Thank you very much,

[kcrisman]

I've been very impressed with Sage and I'm downloading the .dmg for my mac at this moment. 

Thanks!

 

I'm writing here mainly for two reason: 

- firstly because I would like to understand if and how I can set my current python directory as default; 

Sage is an "all-in-one" ecosystem.  There are things like sage-on-gentoo which try to be more modular, but in order to "just work" on the widest variety of systems, Sage uses its own Python.  So if you mean that you want to use your own Python, that is not currently possible on the vanilla Sage.

 

- secondly because I'd like to know If Sage or some of the python libraries installed have a direct support for bayesian econometrics. I'm thinking to use Sage to the writing of my master's thesis and I need to perform estimates with Bayesian VARs using noninformative and informative priors. If there is not a package for bayesian econometrics I would do one myself and I would happy to post it here if you need it.

I don't know much about this at all, but Sage includes R, and R has (as one would expect) a Bayesian microeconomics package:

http://cran.r-project.org/web/packages/bayesm/index.html 

This would be the first place I look, to be honest.  If you need to do further work combining this with Sage, that is very possible due to the back-and-forth between them, though it is less full with R than with Python libraries, of course.

That said, pandas and PyMC (https://pypi.python.org/pypi/pymc/) seem likely candidates for Python, and a good integration of this with Sage is definitely possible, and we would welcome contributions toward this.  I believe many people already use pandas with Sage...

Good luck!

- kcrisman

[Filippo Pellegrino]

Thank you!

I will let you know as soon as possible about pymc, I think that this library does only a part of the work (MCMC sampling) that I need to do. It probably doesn’t include something called “Minnesota priors” and the implementation of a prior using “Dummy observations” for “Vector Autoregressions (VARs)". It is not a true problem, I wrote something for Matlab and I could convert it in Python quite quickly.

I have one last question, is there anyway possible to specify the Sage Python installation path?