Merge pull request #382 from DoubleML/sk-update-sklearn-1.8

Update dependencies  and tests for slearn 1.8.0

Author: SvenKlaassen

doubleml/plm/tests/test_lplr_exceptions.py

@@ -241,7 +241,7 @@ def test_lplr_exception_learner():
 def test_lplr_exception_and_warning_learner():
     # invalid ml_M (must be a classifier with predict_proba)
     with pytest.raises(TypeError):
-        _ = DoubleMLLPLR(dml_data, _DummyNoClassifier(), ml_t, ml_m)
+        _ = DoubleMLLPLR(dml_data, Lasso(), ml_t, ml_m)
     msg = "Invalid learner provided for ml_M: " + r"Lasso\(\) has no method .predict_proba\(\)."
     with pytest.raises(TypeError, match=msg):
         _ = DoubleMLLPLR(dml_data, Lasso(), ml_t, ml_m)

doubleml/tests/test_optuna_multi_wrappers.py

@@ -23,7 +23,7 @@ def _build_apos_object():
     dml_data = dml.DoubleMLData(df, "y", "d")
 
     ml_g = LinearRegression()
-    ml_m = LogisticRegression(max_iter=200, multi_class="auto")
+    ml_m = LogisticRegression(max_iter=200)
 
     return dml.DoubleMLAPOS(
         dml_data,
@@ -38,7 +38,7 @@ def _build_apos_object():
 def _build_qte_object():
     np.random.seed(3141)
     dml_data = make_irm_data(n_obs=80, dim_x=5)
-    ml = LogisticRegression(max_iter=200, multi_class="auto")
+    ml = LogisticRegression(max_iter=200)
 
     return dml.DoubleMLQTE(
         dml_data,
@@ -57,7 +57,7 @@ def _build_did_multi_object():
     dml_panel = DoubleMLPanelData(df, y_col="y", d_cols="d", t_col="t", id_col="id", x_cols=x_cols)
 
     ml_g = LinearRegression()
-    ml_m = LogisticRegression(max_iter=200, multi_class="auto")
+    ml_m = LogisticRegression(max_iter=200)
 
     return dml.did.DoubleMLDIDMulti(
         obj_dml_data=dml_panel,

pyproject.toml

@@ -46,7 +46,7 @@ dev = [
     "pytest>=8.3.0",
     "pytest-cov>=6.0.0",
     "xgboost>=2.1.0",
-    "lightgbm>=4.5.0",
+    "lightgbm>=4.6.0",
     "black>=25.1.0",
     "ruff>=0.11.1",
     "pre-commit>=4.2.0",