BayesBoom 0.1.34

Tools for Bayesian modeling.

Boom stands for 'Bayesian object oriented modeling'.
    It is also the sound your computer makes when it crashes.

    The main part of the Boom library is formulated in terms of abstractions
    for Model, Data, Params, and PosteriorSampler.  A Model is primarily an
    environment where parameters can be learned from data.  The primary
    learning method is Markov chain Monte Carlo, with custom samplers defined
    for specific models.

    The archetypal Boom program looks something like this:

    import BayesBoom as Boom

    some_data = 3 * np.random.randn(100) + 7
    model = Boom.GaussianModel()
    model.set_data(some_data)
    precision_prior = Boom.GammaModel(0.5, 1.5)
    mean_prior = Boom.GaussianModel(0, 10**2)
    poseterior_sampler = Boom.GaussianSemiconjugateSampler(
        model, mean_prior, precision_prior)
    model.set_method(poseterior_sampler)
    niter = 100
    mean_draws = np.zeros(niter)
    sd_draws = np.zeros(niter)
    for i in range(100):
        model.sample_posterior()
        mean_draws[i] = model.mu()
        sd_draws[i] = model.sigma()