Solved – WinBUGs – Code for time series models

arimabayesianregressionwinbugs

I am looking to get started using time series (ARIMA, regression, etc.) evaluating logarithmic stock returns in WinBUGs.

Does anyone have a simple code to get me jump started?

Best Answer

Congdon covers time series models (and WinBUGS code) in a chapter in his Applied Bayesian Modelling book.

Alternatively, BUGS code for some time series models can be produced in my tsbugs package for R (and then run in BUGS via R2WinBUGS or similar such packages). Some of these are applied to the log of stock returns in the example code.

Related Solutions

Solved – Truncated normal distribution with WinBUGS

In openbugs (so I would assume the same in winbugs, but could be wrong) you can do a truncated distribution using the T function, something like:

y[i] ~ dnorm(x[i],tau)T(0,10)

Solved – How to get prediction for a specific variable in WinBUGS

Just add the variable h to the list of the parameters to be monitored. If you are using package like R2WinBUGS, then add variable h to to the list passed to parameters.to.save argument to the bugs function. Then look at your last value in h (the one with NA) - you will get a posterior distribution there.

This is usual way to make predictions in bayesian inference (see also this question). It is nice and simple! No more separation of parameter evaluation and prediction. Everything is done at once. The posterior distrubution of parameters is given by the actual data and propagated to the NA values (as "predictions").

Best Answer

Related Solutions

Solved – Truncated normal distribution with WinBUGS

Solved – How to get prediction for a specific variable in WinBUGS

Related Question