'pybvar' is a package for bayesian vector autoregression in Python. This package is similar to bvars.
This readme contains some examples on the usage of the package.
The package is in a very preliminary stage of its development.
Let's say that we would like to estimate a bayesian VAR with an uninformative prior.
The following code sets up the uninformative prior for a VAR model with lag p=2 and an intercept
prior = uninformative(data,2,True)
In the next step we have to create a bvar object and pass the prior to it. This is done using the following code:
bv = bvar(data,prior)
To start the mcmc algorithm with 10,000 draws and 5,000 burn-in draws and we only want to keep every 5th draw we have to use the following code
results = bv.mcmc(10000,5000,5)