Bayes-Updating with Uncertain Prior Precision

Discussion in 'Math Research' started by Lones.Smith, Mar 20, 2007.

  1. Lones.Smith

    Lones.Smith Guest

    Let's start with a standard model with Gaussian signals (known
    precision s, unknown mean mu) and a Gaussian prior distribution on
    mu (known precision t, and mean m). Fine, done. Now let's forget the
    sample size that went into creating the current posterior m. Instead,
    let's assume that we had a geometric size sample, with sample of size
    n with chance a^n(1-a), for some 0<a<1. Is there then an implied
    conjugate family to work within, perhaps with an unknown prior
    precision on a normal-like variable? DeGroot fails me.

    Many thanks, Lones
    Lones.Smith, Mar 20, 2007
    1. Advertisements

  2. Lones.Smith

    Herman Rubin Guest

    To specify the posterior, one needs to specify the current
    mean AND the current variance. It makes no difference how
    they got those values.
    Herman Rubin, Mar 21, 2007
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.