search

steve-the-bayesian / BOOM

A C++ library for Bayesian modeling, mainly through Markov chain Monte Carlo, but with a few other methods supported. BOOM = "Bayesian Object Oriented Modeling". It is also the sound your computer makes when it crashes.
GNU Lesser General Public License v2.1
36 stars 15 forks source link

Prior Inclusion Probabilities Issue #15

Closed Chyvan closed 6 years ago

Chyvan commented 6 years ago

Currently working with the BSTS package.

I have 120 predictors and define the expected model size as 5. As the prior inclusion probability is defined as expected model size/predictors, the prior inclusion probability should be 0.416. But when check it in the stored bsts model, I can see that it is 0.413. This equals to 5/121. Why is that?

steve-the-bayesian commented 6 years ago

Most likely because an intercept is being added. You can get rid of it by adding a "-1" to your model formula, as in " y ~ . - 1".

On Mon, Mar 26, 2018 at 11:10 AM, Chyvan notifications@github.com wrote:

Currently working with the BSTS package.

I have 120 predictors and define the expected model size as 5. As the prior inclusion probability is defined as expected model size/predictors, the prior inclusion probability should be 0.416. But when check it in the stored bsts model, I can see that it is 0.413. This equals to 5/121. Why is that?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/steve-the-bayesian/BOOM/issues/15, or mute the thread https://github.com/notifications/unsubscribe-auth/AFlR1Ta8Y_XrGhpO2hHA6W9BxIctkGcQks5tiS8ZgaJpZM4S7qFc .

Chyvan commented 6 years ago

ok, thank you for the fast reply!

I have some further questions concerning the package bsts and the method explained in the paper Predicting the Present with BSTS(2014) and Bayesian Variable Selection for Nowcasting Economic Time Series(2015). Is there some other possibility to contact you?