def evaluate_log_predictive_density(model, model_trace_posterior, baseball_dataset):
"""
Evaluate the log probability density of observing the unseen data (season hits)
given a model and empirical distribution over the parameters.
"""
_, test, player_names = train_test_split(baseball_dataset)
at_bats_season, hits_season = test[:, 0], test[:, 1]
test_eval = TracePredictive(conditioned_model,
model_trace_posterior,
num_samples=args.num_samples)
test_eval.run(model, at_bats_season, hits_season)
trace_log_pdf = []
for tr in test_eval.exec_traces:
trace_log_pdf.append(tr.log_prob_sum())
# Use LogSumExp trick to evaluate $log(1/num_samples \sum_i p(new_data | \theta^{i})) $,
# where $\theta^{i}$ are parameter samples from the model's posterior.
posterior_pred_density = log_sum_exp(torch.stack(trace_log_pdf)) - math.log(len(trace_log_pdf))
logging.info("\nLog posterior predictive density")
logging.info("---------------------------------")
logging.info("{:.4f}\n".format(posterior_pred_density))
def evaluate_log_predictive_density(model, model_trace_posterior, baseball_dataset):
"""
Evaluate the log probability density of observing the unseen data (season hits)
given a model and empirical distribution over the parameters.
"""
_, test, player_names = train_test_split(baseball_dataset)
at_bats_season, hits_season = test[:, 0], test[:, 1]
test_eval = TracePredictive(conditioned_model,
model_trace_posterior,
num_samples=args.num_samples)
test_eval.run(model, at_bats_season, hits_season)
trace_log_pdf = []
for tr in test_eval.exec_traces:
trace_log_pdf.append(tr.log_prob_sum())
# Use LogSumExp trick to evaluate $log(1/num_samples \sum_i p(new_data | \theta^{i})) $,
# where $\theta^{i}$ are parameter samples from the model's posterior.
posterior_pred_density = log_sum_exp(torch.stack(trace_log_pdf)) - math.log(len(trace_log_pdf))
logging.info("\nLog posterior predictive density")
logging.info("---------------------------------")
logging.info("{:.4f}\n".format(posterior_pred_density))
def sampling_prediction(self, svi, x_train, y_train, x_test,
num_samples = 1000):
posterior = svi.run(x_train, y_train)
trace_pred = TracePredictive(self.wrapped_model, posterior,
num_samples = num_samples)
post_pred = trace_pred.run(x_test, None)
sites= ['prediction', 'obs']
marginal = get_marginal(post_pred, sites)
site_stats = {}
for i in range(marginal.shape[1]):
site_name = sites[i]
marginal_site = pd.DataFrame(marginal[:, i]).transpose()
site_stats[site_name] = marginal_site.apply(pd.Series.describe,
axis=1)[["mean", "std"]]
mu = site_stats["prediction"]
y_o = site_stats["obs"]
return mu["mean"], mu["std"], y_o["mean"], y_o["std"]
for site, values in summary(posterior, sites).items():
print("Site: {}".format(site))
print(values, "\n")
def wrapped_model(x_data, y_data):
pyro.sample("prediction", dist.Delta(model(x_data, y_data)))
# posterior predictive distribution we can get samples from
trace_pred = TracePredictive(wrapped_model,
posterior,
num_samples=1000)
post_pred = trace_pred.run(x_data, None)
post_summary = summary(post_pred, sites= ['prediction', 'obs'])
mu = post_summary["prediction"]
y = post_summary["obs"]
print("sample y data:")
print(y.head(10))
predictions = pd.DataFrame({
"cont_africa": x_data[:, 0],
"rugged": x_data[:, 1],
"mu_mean": mu["mean"],
"mu_perc_5": mu["5%"],
"mu_perc_95": mu["95%"],
"y_mean": y["mean"],
"y_perc_5": y["5%"],
sites = ["a", "bA", "bR", "bAR", "sigma"]
for site, values in summary(posterior, sites).items():
print("Site: {}".format(site))
print(values, "\n")
def wrapped_model(is_cont_africa, ruggedness, log_gdp):
pyro.sample("prediction", Delta(model(is_cont_africa, ruggedness,
log_gdp)))
# posterior predictive distribution we can get samples from
trace_pred = TracePredictive(wrapped_model, posterior, num_samples=1000)
post_pred = trace_pred.run(is_cont_africa, ruggedness, None)
post_summary = summary(post_pred, sites=['prediction', 'obs'])
mu = post_summary["prediction"]
y = post_summary["obs"]
print("sample y data:")
print(y.head(10))
predictions = pd.DataFrame({
"cont_africa": x_data[:, 0],
"rugged": x_data[:, 1],
"mu_mean": mu["mean"],
"mu_perc_5": mu["5%"],
"mu_perc_95": mu["95%"],
"y_mean": y["mean"],
"y_perc_5": y["5%"],