Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

BetaGeoBetaBinomModel Testing Speedup #1039

Merged
merged 1 commit into from
Sep 15, 2024
Merged

BetaGeoBetaBinomModel Testing Speedup #1039

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
52 changes: 26 additions & 26 deletions tests/clv/models/test_beta_geo_beta_binom.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,13 +40,10 @@ def setup_class(cls):
cls.gamma_true = 0.6567

# Use Quickstart dataset (the CDNOW_sample research data) for testing
test_data = pd.read_csv("data/bgbb_donations.csv")
cls.data = pd.read_csv("data/bgbb_donations.csv")

cls.data = test_data
# cls.customer_id = test_data["customer_id"]
# cls.frequency = test_data["frequency"]
# cls.recency = test_data["recency"]
# cls.T = test_data["T"]
# sample from full dataset for tests involving model fits
cls.sample_data = cls.data.sample(n=1000, random_state=45)

# take sample of all unique recency/frequency/T combinations to test predictive methods
test_customer_ids = [
Expand Down Expand Up @@ -74,8 +71,8 @@ def setup_class(cls):
11103,
]

cls.sample_data = test_data.query("customer_id.isin(@test_customer_ids)")
cls.sample_data_N = len(test_customer_ids)
cls.pred_data = cls.data.query("customer_id.isin(@test_customer_ids)")
cls.pred_data_N = len(test_customer_ids)

# Instantiate model with CDNOW data for testing
cls.model = BetaGeoBetaBinomModel(cls.data)
Expand Down Expand Up @@ -278,13 +275,16 @@ def test_model_repr(self, custom_config):
@pytest.mark.parametrize(
"fit_method, rtol",
[
("mcmc", 0.1),
(
"mcmc",
0.3,
), # higher rtol required for sample_data; within .1 tolerance for full dataset;
("map", 0.2),
],
)
def test_model_convergence(self, fit_method, rtol, model_config):
model = BetaGeoBetaBinomModel(
data=self.data,
data=self.sample_data,
model_config=model_config,
)
model.build_model()
Expand All @@ -307,7 +307,7 @@ def test_model_convergence(self, fit_method, rtol, model_config):
)

def test_fit_result_without_fit(self, model_config):
model = BetaGeoBetaBinomModel(data=self.data, model_config=model_config)
model = BetaGeoBetaBinomModel(data=self.pred_data, model_config=model_config)
with pytest.raises(RuntimeError, match="The model hasn't been fit yet"):
model.fit_result

Expand All @@ -327,20 +327,20 @@ def test_expected_purchases(self, test_t):
true_purchases = (
self.lifetimes_model.conditional_expected_number_of_purchases_up_to_time(
m_periods_in_future=test_t,
frequency=self.sample_data["frequency"],
recency=self.sample_data["recency"],
n_periods=self.sample_data["T"],
frequency=self.pred_data["frequency"],
recency=self.pred_data["recency"],
n_periods=self.pred_data["T"],
)
)

# test parametrization with default data has different dims
est_num_purchases = self.model.expected_purchases(future_t=test_t)
assert est_num_purchases.shape == (self.chains, self.draws, self.N)

data = self.sample_data.assign(future_t=test_t)
data = self.pred_data.assign(future_t=test_t)
est_num_purchases = self.model.expected_purchases(data)

assert est_num_purchases.shape == (self.chains, self.draws, self.sample_data_N)
assert est_num_purchases.shape == (self.chains, self.draws, self.pred_data_N)
assert est_num_purchases.dims == ("chain", "draw", "customer_id")

np.testing.assert_allclose(
Expand Down Expand Up @@ -398,33 +398,33 @@ def test_expected_purchases_new_customer(self):
def test_expected_probability_alive(self, test_t):
true_prob_alive = self.lifetimes_model.conditional_probability_alive(
m_periods_in_future=test_t,
frequency=self.sample_data["frequency"],
recency=self.sample_data["recency"],
n_periods=self.sample_data["T"],
frequency=self.pred_data["frequency"],
recency=self.pred_data["recency"],
n_periods=self.pred_data["T"],
)

# test parametrization with default data has different dims
est_prob_alive = self.model.expected_probability_alive(future_t=test_t)
assert est_prob_alive.shape == (self.chains, self.draws, self.N)

sample_data = self.sample_data.assign(future_t=test_t)
est_prob_alive = self.model.expected_probability_alive(sample_data)
pred_data = self.pred_data.assign(future_t=test_t)
est_prob_alive = self.model.expected_probability_alive(pred_data)

assert est_prob_alive.shape == (self.chains, self.draws, self.sample_data_N)
assert est_prob_alive.shape == (self.chains, self.draws, self.pred_data_N)
assert est_prob_alive.dims == ("chain", "draw", "customer_id")
np.testing.assert_allclose(
true_prob_alive,
est_prob_alive.mean(("chain", "draw")),
rtol=0.01,
)

alt_data = self.sample_data.assign(future_t=7.5)
alt_data = self.pred_data.assign(future_t=7.5)
est_prob_alive_t = self.model.expected_probability_alive(alt_data)
assert est_prob_alive.mean() > est_prob_alive_t.mean()

def test_distribution_new_customer(self) -> None:
mock_model = BetaGeoBetaBinomModel(
data=self.data,
data=self.sample_data,
)
mock_model.build_model()
mock_model.idata = az.from_dict(
Expand All @@ -444,7 +444,7 @@ def test_distribution_new_customer(self) -> None:
random_seed=rng
)
customer_rec_freq = mock_model.distribution_new_customer_recency_frequency(
self.data, T=self.data["T"], random_seed=rng
self.sample_data, T=self.sample_data["T"], random_seed=rng
)
customer_rec = customer_rec_freq.sel(obs_var="recency")
customer_freq = customer_rec_freq.sel(obs_var="frequency")
Expand All @@ -463,7 +463,7 @@ def test_distribution_new_customer(self) -> None:
beta=self.beta_true,
delta=self.delta_true,
gamma=self.gamma_true,
T=self.data["T"],
T=self.sample_data["T"],
),
random_seed=rng,
).T
Expand Down
Loading