示例#1
0
文件: utils.py 项目: dimarkov/pybefit
def log_pred_density(model, samples, *args, **kwargs):
    # waic score of posterior samples
    log_lk = log_likelihood(model, samples, *args, **kwargs)['y']
    ll = log_lk.sum(-1)

    S = ll.shape[0]
    lppd = nn.logsumexp(ll, 0) - jnp.log(S)
    p_waic = jnp.var(ll, axis=0, ddof=1)
    return lppd - p_waic, log_lk
示例#2
0
文件: baseball.py 项目: ucals/numpyro
def predict(model, at_bats, hits, z, rng_key, player_names, train=True):
    header = model.__name__ + (' - TRAIN' if train else ' - TEST')
    predictions = Predictive(model, posterior_samples=z)(rng_key, at_bats)['obs']
    print_results('=' * 30 + header + '=' * 30,
                  predictions,
                  player_names,
                  at_bats,
                  hits)
    if not train:
        post_loglik = log_likelihood(model, z, at_bats, hits)['obs']
        # computes expected log predictive density at each data point
        exp_log_density = logsumexp(post_loglik, axis=0) - jnp.log(jnp.shape(post_loglik)[0])
        # reports log predictive density of all test points
        print('\nLog pointwise predictive density: {:.2f}\n'.format(exp_log_density.sum()))
示例#3
0
def _compute_log_likelihood_null(posterior_samples, data):
    return log_likelihood(model_null, posterior_samples, **data)["obs"]