Skip to content

Fix n_jobs not propagated to auto first-stage learners #1045

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

Open
genrichez wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Fix n_jobs not propagated to auto first-stage learners #1045

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1630d29
Fix n_jobs not propagated to auto first-stage learners
Jul 5, 2026
7d6afa0
Merge branch 'main' into fix/propagate-n-jobs-to-first-stage
genrichez Aug 11, 2026
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
6 changes: 4 additions & 2 deletions econml/dml/causal_forest.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -683,10 +683,12 @@ def _gen_featurizer(self):
return clone(self.featurizer, safe=False)

def _gen_model_y(self):
return _make_first_stage_selector(self.model_y, self.discrete_outcome, self.random_state)
return _make_first_stage_selector(self.model_y, self.discrete_outcome, self.random_state,
n_jobs=self.n_jobs)

def _gen_model_t(self):
return _make_first_stage_selector(self.model_t, self.discrete_treatment, self.random_state)
return _make_first_stage_selector(self.model_t, self.discrete_treatment, self.random_state,
n_jobs=self.n_jobs)

def _gen_model_final(self):
return MultiOutputGRF(CausalForest(n_estimators=self.n_estimators,
Expand Down
17 changes: 11 additions & 6 deletions econml/dml/dml.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -120,12 +120,13 @@ def best_score(self):
return self._model.best_score


def _make_first_stage_selector(model, is_discrete, random_state):
def _make_first_stage_selector(model, is_discrete, random_state, n_jobs=None):
if model == 'auto':
model = ['forest', 'linear']
return _FirstStageSelector(get_selector(model,
is_discrete=is_discrete,
random_state=random_state),
random_state=random_state,
n_jobs=n_jobs),
discrete_target=is_discrete)


Expand Down Expand Up @@ -561,10 +562,12 @@ def _gen_featurizer(self):
return clone(self.featurizer, safe=False)

def _gen_model_y(self):
return _make_first_stage_selector(self.model_y, self.discrete_outcome, self.random_state)
return _make_first_stage_selector(self.model_y, self.discrete_outcome, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t(self):
return _make_first_stage_selector(self.model_t, self.discrete_treatment, self.random_state)
return _make_first_stage_selector(self.model_t, self.discrete_treatment, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_final(self):
return clone(self.model_final, safe=False)
Expand Down Expand Up @@ -1647,11 +1650,13 @@ def _gen_featurizer(self):

def _gen_model_y(self):
return _make_first_stage_selector(self.model_y, is_discrete=self.discrete_outcome,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t(self):
return _make_first_stage_selector(self.model_t, is_discrete=self.discrete_treatment,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_final(self):
return clone(self.model_final, safe=False)
Expand Down
10 changes: 6 additions & 4 deletions econml/dr/_drlearner.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -189,10 +189,10 @@ def predict(self, Y, T, X=None, W=None, *, sample_weight=None, groups=None):
return Y_pred.reshape(Y.shape + (T.shape[1] + 1,)), propensities, raw_propensities


def _make_first_stage_selector(model, is_discrete, random_state):
def _make_first_stage_selector(model, is_discrete, random_state, n_jobs=None):
if model == "auto":
model = ['linear', 'forest']
return get_selector(model, is_discrete=is_discrete, random_state=random_state)
return get_selector(model, is_discrete=is_discrete, random_state=random_state, n_jobs=n_jobs)


class _ModelFinal:
Expand Down Expand Up @@ -678,8 +678,10 @@ def _get_inference_options(self):
return options

def _gen_ortho_learner_model_nuisance(self):
model_propensity = _make_first_stage_selector(self.model_propensity, True, self.random_state)
model_regression = _make_first_stage_selector(self.model_regression, self.discrete_outcome, self.random_state)
model_propensity = _make_first_stage_selector(self.model_propensity, True, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))
model_regression = _make_first_stage_selector(self.model_regression, self.discrete_outcome, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

return _ModelNuisance(model_propensity, model_regression, self.min_propensity, self.discrete_outcome)

Expand Down
31 changes: 21 additions & 10 deletions econml/iv/dml/_dml.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -409,26 +409,31 @@ def _gen_ortho_learner_model_final(self):
return _OrthoIVModelFinal(self._gen_model_final(), self._gen_featurizer(), self.fit_cate_intercept)

def _gen_ortho_learner_model_nuisance(self):
n_jobs = getattr(self, 'n_jobs', None)
model_y = _make_first_stage_selector(self.model_y_xw,
is_discrete=self.discrete_outcome,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

model_t = _make_first_stage_selector(self.model_t_xw,
is_discrete=self.discrete_treatment,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

if self.projection:
# train E[T|X,W,Z]
model_z = _make_first_stage_selector(self.model_t_xwz,
is_discrete=self.discrete_treatment,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

else:
# train E[Z|X,W]
# note: discrete_instrument rather than discrete_treatment in call to _make_first_stage_selector
model_z = _make_first_stage_selector(self.model_z_xw,
is_discrete=self.discrete_instrument,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

return _OrthoIVNuisanceSelector(model_y, model_t, model_z,
self.projection)
Expand Down Expand Up @@ -1189,13 +1194,16 @@ def _gen_featurizer(self):
return clone(self.featurizer, safe=False)

def _gen_model_y_xw(self):
return _make_first_stage_selector(self.model_y_xw, self.discrete_outcome, self.random_state)
return _make_first_stage_selector(self.model_y_xw, self.discrete_outcome, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t_xw(self):
return _make_first_stage_selector(self.model_t_xw, self.discrete_treatment, self.random_state)
return _make_first_stage_selector(self.model_t_xw, self.discrete_treatment, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t_xwz(self):
return _make_first_stage_selector(self.model_t_xwz, self.discrete_treatment, self.random_state)
return _make_first_stage_selector(self.model_t_xwz, self.discrete_treatment, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_final(self):
return clone(self.model_final, safe=False)
Expand Down Expand Up @@ -1576,13 +1584,16 @@ def _gen_featurizer(self):
return clone(self.featurizer, safe=False)

def _gen_model_y_xw(self):
return _make_first_stage_selector(self.model_y_xw, self.discrete_outcome, self.random_state)
return _make_first_stage_selector(self.model_y_xw, self.discrete_outcome, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t_xw(self):
return _make_first_stage_selector(self.model_t_xw, self.discrete_treatment, self.random_state)
return _make_first_stage_selector(self.model_t_xw, self.discrete_treatment, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t_xwz(self):
return _make_first_stage_selector(self.model_t_xwz, self.discrete_treatment, self.random_state)
return _make_first_stage_selector(self.model_t_xwz, self.discrete_treatment, self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_final(self):
return clone(self.model_final, safe=False)
Expand Down
29 changes: 20 additions & 9 deletions econml/iv/dr/_dr.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -663,30 +663,37 @@ def _gen_prel_model_effect(self):
return clone(self.prel_model_effect, safe=False)

def _gen_ortho_learner_model_nuisance(self):
model_y_xw = _make_first_stage_selector(self.model_y_xw, self.discrete_outcome, self.random_state)
model_t_xw = _make_first_stage_selector(self.model_t_xw, self.discrete_treatment, self.random_state)
n_jobs = getattr(self, 'n_jobs', None)
model_y_xw = _make_first_stage_selector(self.model_y_xw, self.discrete_outcome, self.random_state,
n_jobs=n_jobs)
model_t_xw = _make_first_stage_selector(self.model_t_xw, self.discrete_treatment, self.random_state,
n_jobs=n_jobs)

if self.projection:
# this is a regression model since the instrument E[T|X,W,Z] is always continuous
model_tz_xw = _make_first_stage_selector(self.model_tz_xw,
is_discrete=False,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

# we're using E[T|X,W,Z] as the instrument
model_z = _make_first_stage_selector(self.model_t_xwz,
is_discrete=self.discrete_treatment,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

else:
model_tz_xw = _make_first_stage_selector(self.model_tz_xw,
is_discrete=(self.discrete_treatment and
self.discrete_instrument and
not self.fit_cov_directly),
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

model_z = _make_first_stage_selector(self.model_z_xw,
is_discrete=self.discrete_instrument,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)

return [_BaseDRIVNuisanceSelector(prel_model_effect=self._gen_prel_model_effect(),
model_y_xw=model_y_xw,
Expand Down Expand Up @@ -2517,10 +2524,13 @@ def _gen_prel_model_effect(self):
return clone(self.prel_model_effect, safe=False)

def _gen_ortho_learner_model_nuisance(self):
n_jobs = getattr(self, 'n_jobs', None)
model_y_xw = _make_first_stage_selector(self.model_y_xw,
is_discrete=self.discrete_outcome,
random_state=self.random_state)
model_t_xwz = _make_first_stage_selector(self.model_t_xwz, is_discrete=True, random_state=self.random_state)
random_state=self.random_state,
n_jobs=n_jobs)
model_t_xwz = _make_first_stage_selector(self.model_t_xwz, is_discrete=True,
random_state=self.random_state, n_jobs=n_jobs)

if self.z_propensity == "auto":
dummy_z = DummyClassifier(strategy="prior")
Expand All @@ -2529,7 +2539,8 @@ def _gen_ortho_learner_model_nuisance(self):
else:
raise ValueError("Only 'auto' or float is allowed!")

dummy_z = _make_first_stage_selector(dummy_z, is_discrete=True, random_state=self.random_state)
dummy_z = _make_first_stage_selector(dummy_z, is_discrete=True, random_state=self.random_state,
n_jobs=n_jobs)

return _IntentToTreatDRIVNuisanceSelector(model_y_xw, model_t_xwz, dummy_z, self._gen_prel_model_effect())

Expand Down
6 changes: 4 additions & 2 deletions econml/panel/dml/_dml.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -576,12 +576,14 @@ def _gen_featurizer(self):
def _gen_model_y(self):
return _make_first_stage_selector(self.model_y,
is_discrete=self.discrete_outcome,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_t(self):
return _make_first_stage_selector(self.model_t,
is_discrete=self.discrete_treatment,
random_state=self.random_state)
random_state=self.random_state,
n_jobs=getattr(self, 'n_jobs', None))

def _gen_model_final(self):
return StatsModelsLinearRegression(fit_intercept=False)
Expand Down
21 changes: 11 additions & 10 deletions econml/sklearn_extensions/model_selection.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -616,37 +616,38 @@ def best_score(self):
return self._best_score


def get_selector(input, is_discrete, *, random_state=None, cv=None, wrapper=GridSearchCV, needs_scoring=False):
def get_selector(input, is_discrete, *, random_state=None, cv=None, n_jobs=None,
wrapper=GridSearchCV, needs_scoring=False):
named_models = {
'linear': (LogisticRegressionCV(random_state=random_state, cv=cv) if is_discrete
else WeightedLassoCVWrapper(random_state=random_state, cv=cv)),
else WeightedLassoCVWrapper(random_state=random_state, cv=cv, n_jobs=n_jobs)),
'poly': ([make_pipeline(PolynomialFeatures(d),
(LogisticRegressionCV(random_state=random_state, cv=cv) if is_discrete
else WeightedLassoCVWrapper(random_state=random_state, cv=cv)))
else WeightedLassoCVWrapper(random_state=random_state, cv=cv, n_jobs=n_jobs)))
for d in range(1, 4)]),
'forest': (GridSearchCV(RandomForestClassifier(random_state=random_state) if is_discrete
else RandomForestRegressor(random_state=random_state),
param_grid={}, cv=cv)),
'forest': (GridSearchCV(RandomForestClassifier(n_jobs=n_jobs, random_state=random_state) if is_discrete
else RandomForestRegressor(n_jobs=n_jobs, random_state=random_state),
param_grid={}, cv=cv, n_jobs=n_jobs)),
'gbf': (GridSearchCV(GradientBoostingClassifier(random_state=random_state) if is_discrete
else GradientBoostingRegressor(random_state=random_state),
param_grid={}, cv=cv)),
param_grid={}, cv=cv, n_jobs=n_jobs)),
'nnet': (GridSearchCV(MLPClassifier(random_state=random_state) if is_discrete
else MLPRegressor(random_state=random_state),
param_grid={}, cv=cv)),
param_grid={}, cv=cv, n_jobs=n_jobs)),
'automl': ["poly", "forest", "gbf", "nnet"],
}
if isinstance(input, ModelSelector): # we've already got a model selector, don't need to do anything
return input
elif isinstance(input, list): # we've got a list; call get_selector on each element, then wrap in a ListSelector
models = [get_selector(model, is_discrete,
random_state=random_state, cv=cv, wrapper=wrapper,
random_state=random_state, cv=cv, n_jobs=n_jobs, wrapper=wrapper,
needs_scoring=True) # we need to score to compare outputs to each other
for model in input]
return ListSelector(models)
elif isinstance(input, str): # we've got a string; look it up
if input in named_models:
return get_selector(named_models[input], is_discrete,
random_state=random_state, cv=cv, wrapper=wrapper,
random_state=random_state, cv=cv, n_jobs=n_jobs, wrapper=wrapper,
needs_scoring=needs_scoring)
else:
raise ValueError(f"Unknown model type: {input}, must be one of {named_models.keys()}")
Expand Down