def test_smoke(num_data, feature_dim, outcome_dist):
x, t, y = generate_data(num_data, feature_dim)
if outcome_dist == "exponential":
y.clamp_(min=1e-20)
cevae = CEVAE(feature_dim, outcome_dist)
cevae.fit(x, t, y, num_epochs=2)
ite = cevae.ite(x)
assert ite.shape == (num_data,)
def fit(self, X, treatment, y, p=None):
"""
Fits CEVAE.
Args:
X (np.matrix or np.array or pd.Dataframe): a feature matrix
treatment (np.array or pd.Series): a treatment vector
y (np.array or pd.Series): an outcome vector
"""
X, treatment, y = convert_pd_to_np(X, treatment, y)
self.cevae = CEVAEModel(outcome_dist=self.outcome_dist,
feature_dim=X.shape[-1],
latent_dim=self.latent_dim,
hidden_dim=self.hidden_dim,
num_layers=self.num_layers)
self.cevae.fit(x=torch.tensor(X, dtype=torch.float),
t=torch.tensor(treatment, dtype=torch.float),
y=torch.tensor(y, dtype=torch.float),
num_epochs=self.num_epochs,
batch_size=self.batch_size,
learning_rate=self.learning_rate,
learning_rate_decay=self.learning_rate_decay,
weight_decay=self.weight_decay)
def main(args):
pyro.enable_validation(__debug__)
if args.cuda:
torch.set_default_tensor_type('torch.cuda.FloatTensor')
# Generate synthetic data.
pyro.set_rng_seed(args.seed)
x_train, t_train, y_train, _ = generate_data(args)
# Train.
pyro.set_rng_seed(args.seed)
pyro.clear_param_store()
cevae = CEVAE(feature_dim=args.feature_dim,
latent_dim=args.latent_dim,
hidden_dim=args.hidden_dim,
num_layers=args.num_layers,
num_samples=10)
cevae.fit(x_train, t_train, y_train,
num_epochs=args.num_epochs,
batch_size=args.batch_size,
learning_rate=args.learning_rate,
learning_rate_decay=args.learning_rate_decay,
weight_decay=args.weight_decay)
# Evaluate.
x_test, t_test, y_test, true_ite = generate_data(args)
true_ate = true_ite.mean()
print("true ATE = {:0.3g}".format(true_ate.item()))
naive_ate = y_test[t_test == 1].mean() - y_test[t_test == 0].mean()
print("naive ATE = {:0.3g}".format(naive_ate))
if args.jit:
cevae = cevae.to_script_module()
est_ite = cevae.ite(x_test)
est_ate = est_ite.mean()
print("estimated ATE = {:0.3g}".format(est_ate.item()))
def test_serialization(jit, feature_dim, outcome_dist):
x, t, y = generate_data(num_data=32, feature_dim=feature_dim)
if outcome_dist == "exponential":
y.clamp_(min=1e-20)
cevae = CEVAE(feature_dim, outcome_dist=outcome_dist, num_samples=1000, hidden_dim=32)
cevae.fit(x, t, y, num_epochs=4, batch_size=8)
pyro.set_rng_seed(0)
expected_ite = cevae.ite(x)
if jit:
traced_cevae = cevae.to_script_module()
f = io.BytesIO()
torch.jit.save(traced_cevae, f)
f.seek(0)
loaded_cevae = torch.jit.load(f)
else:
f = io.BytesIO()
with warnings.catch_warnings():
warnings.filterwarnings("ignore", category=UserWarning)
torch.save(cevae, f)
f.seek(0)
loaded_cevae = torch.load(f)
pyro.set_rng_seed(0)
actual_ite = loaded_cevae.ite(x)
assert_close(actual_ite, expected_ite, atol=0.1)
class CEVAE:
def __init__(self, outcome_dist="studentt", latent_dim=20, hidden_dim=200, num_epochs=50, num_layers=3,
batch_size=100, learning_rate=1e-3, learning_rate_decay=0.1, num_samples=1000, weight_decay=1e-4):
"""
Initializes CEVAE.
Args:
outcome_dist (str): Outcome distribution as one of: "bernoulli" , "exponential", "laplace", "normal",
and "studentt"
latent_dim (int) : Dimension of the latent variable
hidden_dim (int) : Dimension of hidden layers of fully connected networks
num_epochs (int): Number of training epochs
num_layers (int): Number of hidden layers in fully connected networks
batch_size (int): Batch size
learning_rate (int): Learning rate
learning_rate_decay (float/int): Learning rate decay over all epochs; the per-step decay rate will
depend on batch size and number of epochs such that the initial
learning rate will be learning_rate and the
final learning rate will be learning_rate * learning_rate_decay
num_samples (int) : Number of samples to calculate ITE
weight_decay (float) : Weight decay
"""
self.outcome_dist = outcome_dist
self.latent_dim = latent_dim
self.hidden_dim = hidden_dim
self.num_epochs = num_epochs
self.num_layers = num_layers
self.batch_size = batch_size
self.learning_rate = learning_rate
self.learning_rate_decay = learning_rate_decay
self.num_samples = num_samples
self.weight_decay = weight_decay
def fit(self, X, treatment, y, p=None):
"""
Fits CEVAE.
Args:
X (np.matrix or np.array or pd.Dataframe): a feature matrix
treatment (np.array or pd.Series): a treatment vector
y (np.array or pd.Series): an outcome vector
"""
X, treatment, y = convert_pd_to_np(X, treatment, y)
self.cevae = CEVAEModel(outcome_dist=self.outcome_dist,
feature_dim=X.shape[-1],
latent_dim=self.latent_dim,
hidden_dim=self.hidden_dim,
num_layers=self.num_layers)
self.cevae.fit(x=torch.tensor(X, dtype=torch.float),
t=torch.tensor(treatment, dtype=torch.float),
y=torch.tensor(y, dtype=torch.float),
num_epochs=self.num_epochs,
batch_size=self.batch_size,
learning_rate=self.learning_rate,
learning_rate_decay=self.learning_rate_decay,
weight_decay=self.weight_decay)
def predict(self, X, treatment=None, y=None, p=None):
"""
Calls predict on fitted DragonNet.
Args:
X (np.matrix or np.array or pd.Dataframe): a feature matrix
Returns:
(np.ndarray): Predictions of treatment effects.
"""
return self.cevae.ite(torch.tensor(X, dtype=torch.float),
num_samples=self.num_samples,
batch_size=self.batch_size).cpu().numpy()
def fit_predict(self, X, treatment, y, p=None):
"""
Fits the CEVAE model and then predicts.
Args:
X (np.matrix or np.array or pd.Dataframe): a feature matrix
treatment (np.array or pd.Series): a treatment vector
y (np.array or pd.Series): an outcome vector
Returns:
(np.ndarray): Predictions of treatment effects.
"""
self.fit(X, treatment, y)
return self.predict(X)