def test_compare_tmle_continuous(self, cf):
cf['cd4_wk45'] = np.log(cf['cd4_wk45'])
stmle = StochasticTMLE(cf, exposure='art', outcome='cd4_wk45')
stmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.outcome_model(
'art + male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.fit(p=1.0, samples=1)
all_treat = stmle.marginal_outcome
stmle.fit(p=0.0, samples=1)
non_treat = stmle.marginal_outcome
tmle = TMLE(cf, exposure='art', outcome='cd4_wk45')
tmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0',
print_results=False)
tmle.outcome_model(
'art + male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0',
print_results=False)
tmle.fit()
expected = tmle.average_treatment_effect
npt.assert_allclose(expected, all_treat - non_treat, atol=1e-3)
def test_error_summary(self, df):
stmle = StochasticTMLE(df=df, exposure='art', outcome='dead')
stmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.outcome_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
with pytest.raises(ValueError):
stmle.summary()
def test_gmodel_bound(self, simple_df):
# Comparing to SAS Poisson model
sas_preds = [
2, 1 / 0.55, 1 / 0.45, 1 / 0.55, 2, 2, 1 / 0.45, 1 / 0.55, 2
]
stmle = StochasticTMLE(df=simple_df, exposure='A', outcome='C')
stmle.exposure_model('W', bound=[0.45, 0.55])
est_preds = 1 / stmle._denominator_
npt.assert_allclose(sas_preds, est_preds, atol=1e-6)
def test_gmodel_params(self, simple_df):
# Comparing to SAS Poisson model
sas_preds = [
2.0, 1.6666666667, 2.5, 1.6666666667, 2, 2, 2.5, 1.6666666667, 2
]
stmle = StochasticTMLE(df=simple_df, exposure='A', outcome='C')
stmle.exposure_model('W')
est_preds = 1 / stmle._denominator_
npt.assert_allclose(sas_preds, est_preds, atol=1e-6)
def test_error_p_cond_len(self, df):
stmle = StochasticTMLE(df=df, exposure='art', outcome='dead')
stmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.outcome_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
with pytest.raises(ValueError):
stmle.fit(p=[0.1], conditional=["df['male']==1", "df['male']==0"])
with pytest.raises(ValueError):
stmle.fit(p=[0.1, 0.3], conditional=["df['male']==1"])
def test_machine_learning_runs(self, df):
# Only verifies that machine learning doesn't throw an error
log = LogisticRegression(penalty='l1',
solver='liblinear',
random_state=201)
tmle = StochasticTMLE(df, exposure='art', outcome='dead')
tmle.exposure_model(
'male + age0 + cd40 + cd4_rs1 + cd4_rs2 + dvl0 + male:dvl0',
custom_model=log)
tmle.outcome_model('art + male + age0 + dvl0 + cd40',
custom_model=log)
tmle.fit(p=0.4, samples=20)
def test_calculate_epsilon2(self, cf):
stmle = StochasticTMLE(cf, exposure='art', outcome='cd4_wk45')
stmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.outcome_model(
'art + male + age0 + age_rs1 + age_rs2 + dvl0 + cd40 + cd4_rs1 + cd4_rs2'
)
stmle.fit(p=0.15, samples=1)
npt.assert_allclose(-0.0059476590, stmle.epsilon, atol=1e-6)
stmle.fit(p=0.4, samples=1)
npt.assert_allclose(-0.0154923643, stmle.epsilon, atol=1e-6)
def test_calculate_epsilon1(self, df):
stmle = StochasticTMLE(df, exposure='art', outcome='dead')
stmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.outcome_model(
'art + male + age0 + age_rs1 + age_rs2 + dvl0 + cd40 + cd4_rs1 + cd4_rs2'
)
stmle.fit(p=0.15, samples=1)
npt.assert_allclose(-0.0157043107, stmle.epsilon, atol=1e-6)
stmle.fit(p=0.4, samples=1)
npt.assert_allclose(-0.0381559025, stmle.epsilon, atol=1e-6)
def test_compare_tmle_binary(self, df):
stmle = StochasticTMLE(df, exposure='art', outcome='dead')
stmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.outcome_model(
'art + male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0'
)
stmle.fit(p=1.0, samples=1)
all_treat = stmle.marginal_outcome
stmle.fit(p=0.0, samples=1)
non_treat = stmle.marginal_outcome
tmle = TMLE(df, exposure='art', outcome='dead')
tmle.exposure_model(
'male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0',
print_results=False)
tmle.outcome_model(
'art + male + age0 + age_rs1 + age_rs2 + cd40 + cd4_rs1 + cd4_rs2 + dvl0',
print_results=False)
tmle.fit()
expected = tmle.risk_difference
npt.assert_allclose(expected, all_treat - non_treat, atol=1e-4)
for p in prop_treated] + ['var_' + str(p) for p in prop_treated])
########################################
# Running simulation
########################################
for i in range(n_mc):
# Generating Data
H = naloxone_dgm(network=G, restricted=restrict)
df = network_to_df(H)
results.loc[i, 'inc_' + exposure] = np.mean(df[exposure])
results.loc[i, 'inc_' + outcome] = np.mean(df[outcome])
if independent:
# Stochastic TMLE
stmle = StochasticTMLE(df, exposure=exposure, outcome=outcome)
stmle.exposure_model(gi_model, bound=0.01)
stmle.outcome_model(qi_model)
for p in prop_treated: # loops through all treatment plans
try:
if shift:
z = odds_to_probability(np.exp(log_odds + p))
stmle.fit(p=z)
else:
stmle.fit(p=p)
results.loc[i, 'bias_' +
str(p)] = stmle.marginal_outcome - truth[p]
results.loc[i, 'var_' + str(p)] = stmle.conditional_se**2
results.loc[i, 'lcl_' + str(p)] = stmle.conditional_ci[0]
results.loc[i, 'ucl_' + str(p)] = stmle.conditional_ci[1]
except:
results.loc[i, 'bias_' + str(p)] = np.nan
def test_marginal_vector_length_stoch(self, df):
stmle = StochasticTMLE(df=df, exposure='art', outcome='dead')
stmle.exposure_model('male')
stmle.outcome_model('art + male + age0')
stmle.fit(p=0.4, samples=7)
assert len(stmle.marginals_vector) == 7