autogluon.core 1.2.1b20250110__py3-none-any.whl → 1.2.1b20250111__py3-none-any.whl

autogluon/core/models/abstract/_tags.py

@@ -1,6 +1,12 @@
 _DEFAULT_TAGS = {
-    # Whether the model can produce out-of-fold (or similar) predictions of the training data without being significantly overfit.
+    # [Advanced] Whether the model can support fitting on 100% of the data and then getting unbiased predictions on the same data.
+    # it fit on by exploiting special properties of the model architecture.
+    # For example, random forest uses only a portion of the training data randomly for each decision tree.
+    # We can therefore use the out-of-bag predictions to obtain unbiased predictions.
+    # Note that models that specify this as True must implement a `predict_proba_oof` method.
+    # Refer to RandomForestModel or KNeighborsModel for reference implementations.
     "valid_oof": False,
+
     # Whether the model can be refit using the combined train and val data as training and no validation data without issue.
     #  TL;DR: Keep value as False unless you know what you are doing. This is advanced functionality.
     #  If False, when calling predictor.refit_full(), this model will simply be duplicated (if non-bag) or will have the first fold model duplicated (if bag).

autogluon/core/models/ensemble/bagged_ensemble_model.py

@@ -193,19 +193,19 @@ class BaggedEnsembleModel(AbstractModel):
 
     def _fit(
         self,
-        X,
-        y,
-        X_val=None,
-        y_val=None,
-        X_pseudo=None,
-        y_pseudo=None,
-        k_fold=None,
-        k_fold_start=0,
-        k_fold_end=None,
-        n_repeats=1,
-        n_repeat_start=0,
-        groups=None,
-        _skip_oof=False,
+        X: pd.DataFrame,
+        y: pd.Series,
+        X_val: pd.DataFrame = None,
+        y_val: pd.Series = None,
+        X_pseudo: pd.DataFrame = None,
+        y_pseudo: pd.Series = None,
+        k_fold: int = None,
+        k_fold_start: int = 0,
+        k_fold_end: int = None,
+        n_repeats: int = 1,
+        n_repeat_start: int = 0,
+        groups: pd.Series = None,
+        _skip_oof: bool = False,
         **kwargs,
     ):
         use_child_oof = self.params.get("use_child_oof", False)
@@ -223,8 +223,6 @@ class BaggedEnsembleModel(AbstractModel):
             k_fold, k_fold_end = self._update_k_fold(k_fold=k_fold, k_fold_end=k_fold_end)
         if k_fold is None and groups is None:
             k_fold = 5
-        if k_fold is not None and k_fold < 1:
-            k_fold = 1
         if k_fold is None or k_fold > 1:
             k_fold = self._get_cv_splitter(n_splits=k_fold, n_repeats=n_repeats, groups=groups).n_splits
         max_sets = self._get_model_params().get("max_sets", None)
@@ -238,6 +236,7 @@ class BaggedEnsembleModel(AbstractModel):
             n_repeats=n_repeats,
             n_repeat_start=n_repeat_start,
             groups=groups,
+            use_child_oof=use_child_oof,
         )
         if k_fold_end is None:
             k_fold_end = k_fold
@@ -327,7 +326,7 @@ class BaggedEnsembleModel(AbstractModel):
             else:
                 return self
 
-    def _update_k_fold(self, k_fold, k_fold_end=None, verbose=True):
+    def _update_k_fold(self, k_fold: int, k_fold_end: int = None, verbose: bool = True) -> tuple[int, int]:
         """Update k_fold and k_fold_end in case num_folds was specified"""
         k_fold_override = self.params.get("num_folds", None)
         if k_fold_override is not None:
@@ -343,7 +342,17 @@ class BaggedEnsembleModel(AbstractModel):
         assert self.is_initialized(), "Model must be initialized before calling self._get_child_aux_val!"
         return self._params_aux_child.get(key, default)
 
-    def _validate_bag_kwargs(self, *, k_fold, k_fold_start, k_fold_end, n_repeats, n_repeat_start, groups):
+    def _validate_bag_kwargs(
+        self,
+        *,
+        k_fold: int,
+        k_fold_start: int,
+        k_fold_end: int,
+        n_repeats: int,
+        n_repeat_start: int,
+        groups: pd.Series | None,
+        use_child_oof: bool,
+    ):
         if groups is not None:
             if self._n_repeats_finished != 0:
                 raise AssertionError("Bagged models cannot call fit with `groups` specified when a full k-fold set has already been fit.")
@@ -356,7 +365,7 @@ class BaggedEnsembleModel(AbstractModel):
         if k_fold is None:
             raise ValueError("k_fold cannot be None.")
         if k_fold < 1:
-            raise ValueError(f"k_fold must be equal or greater than 1, value: ({k_fold})")
+            raise ValueError(f"k_fold must be equal or greater than 1, value: {k_fold}")
         if n_repeat_start != self._n_repeats_finished:
             raise ValueError(f"n_repeat_start must equal self._n_repeats_finished, values: ({n_repeat_start}, {self._n_repeats_finished})")
         if n_repeats <= n_repeat_start:
@@ -370,7 +379,26 @@ class BaggedEnsembleModel(AbstractModel):
             # TODO: Remove this limitation
             raise ValueError(f"k_fold_end must equal k_fold when (n_repeats - n_repeat_start) > 1, values: ({k_fold_end}, {k_fold})")
         if self._k is not None and self._k != k_fold:
-            raise ValueError(f"k_fold must equal previously fit k_fold value for the current n_repeat, values: (({k_fold}, {self._k})")
+            raise ValueError(f"k_fold must equal previously fit k_fold value for the current n_repeat, values: ({k_fold}, {self._k})")
+        if use_child_oof and not self._get_tags_child().get("valid_oof", False):
+            raise AssertionError(
+                f"`use_child_oof=True` was specified, "
+                f"but the model {self._child_type.__name__} does not support this option. (valid_oof=False)\n"
+                f"\tTo enable this logic, `{self._child_type.__name__}._predict_proba_oof` must be implemented "
+                f"and `tags['valid_oof'] = True` must be set in `{self._child_type.__name__}._more_tags`."
+            )
+        if k_fold == 1 and not use_child_oof and not self._get_tags().get("can_get_oof_from_train", False):
+            logger.log(
+                30,
+                f"\tWARNING: Fitting bagged model with `k_fold=1`, "
+                f"but this model doesn't support getting out-of-fold predictions from training data!\n"
+                f"\t\tThe model will be fit on 100% of the training data without any validation split.\n"
+                f"\t\tIt will then predict on the same data used to train for generating out-of-fold predictions. "
+                f"This will likely be EXTREMELY overfit and produce terrible results.\n"
+                f"\t\tWe strongly recommend not forcing bagged models to use `k_fold=1`. "
+                f"Instead, specify `use_child_oof=True` if the model supports this option."
+            )
+
 
     def predict_proba_children(
         self,
@@ -557,8 +585,9 @@ class BaggedEnsembleModel(AbstractModel):
                     logger.log(
                         30,
                         f"\tWARNING: Setting `self._oof_pred_proba` by predicting on train directly! "
-                        f"This is probably a bug and should be investigated...\n"
-                        f'\tIf this is intended, set the model tag "can_get_oof_from_train" to True '
+                        f"This is probably a bug or the user specified `num_folds=1` "
+                        f"as an `ag_args_ensemble` hyperparameter... Results may be very poor.\n"
+                        f'\t\tIf this is intended, set the model tag "can_get_oof_from_train" to True '
                         f"in `{self.__class__.__name__}._more_tags` to avoid this warning.",
                     )
                 self._oof_pred_proba = model_base.predict_proba(X=X)  # TODO: Cheater value, will be overfit to valid set

autogluon/core/version.py

@@ -1,3 +1,3 @@
 """This is the autogluon version file."""
-__version__ = '1.2.1b20250110'
+__version__ = '1.2.1b20250111'
 __lite__ = False

{autogluon.core-1.2.1b20250110.dist-info → autogluon.core-1.2.1b20250111.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.core
-Version: 1.2.1b20250110
+Version: 1.2.1b20250111
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community
@@ -43,12 +43,12 @@ Requires-Dist: tqdm<5,>=4.38
 Requires-Dist: requests
 Requires-Dist: matplotlib<3.11,>=3.7.0
 Requires-Dist: boto3<2,>=1.10
-Requires-Dist: autogluon.common==1.2.1b20250110
+Requires-Dist: autogluon.common==1.2.1b20250111
 Provides-Extra: all
+Requires-Dist: hyperopt<0.2.8,>=0.2.7; extra == "all"
 Requires-Dist: ray[default,tune]<2.41,>=2.10.0; extra == "all"
 Requires-Dist: ray[default]<2.41,>=2.10.0; extra == "all"
 Requires-Dist: pyarrow>=15.0.0; extra == "all"
-Requires-Dist: hyperopt<0.2.8,>=0.2.7; extra == "all"
 Provides-Extra: ray
 Requires-Dist: ray[default]<2.41,>=2.10.0; extra == "ray"
 Provides-Extra: raytune
@@ -56,11 +56,11 @@ Requires-Dist: pyarrow>=15.0.0; extra == "raytune"
 Requires-Dist: ray[default,tune]<2.41,>=2.10.0; extra == "raytune"
 Requires-Dist: hyperopt<0.2.8,>=0.2.7; extra == "raytune"
 Provides-Extra: tests
-Requires-Dist: types-setuptools; extra == "tests"
-Requires-Dist: pytest-mypy; extra == "tests"
 Requires-Dist: flake8; extra == "tests"
-Requires-Dist: pytest; extra == "tests"
+Requires-Dist: pytest-mypy; extra == "tests"
+Requires-Dist: types-setuptools; extra == "tests"
 Requires-Dist: types-requests; extra == "tests"
+Requires-Dist: pytest; extra == "tests"
 
 
 

{autogluon.core-1.2.1b20250110.dist-info → autogluon.core-1.2.1b20250111.dist-info}/RECORD

@@ -1,9 +1,9 @@
-autogluon.core-1.2.1b20250110-py3.8-nspkg.pth,sha256=cQGwpuGPqg1GXscIwt-7PmME1OnSpD-7ixkikJ31WAY,554
+autogluon.core-1.2.1b20250111-py3.8-nspkg.pth,sha256=cQGwpuGPqg1GXscIwt-7PmME1OnSpD-7ixkikJ31WAY,554
 autogluon/core/__init__.py,sha256=8KfvvHzXX3a4q6z43Dw1yE7VtbAoiSMaglVpKDy6Xeg,245
 autogluon/core/_setup_utils.py,sha256=NqlGK6So0KG5M0LbBJNT1TI3iAmG93kd_6Brih6y2gQ,6935
 autogluon/core/constants.py,sha256=nEVLdSFJ-5O-tz3jUD3qPX65RMp7g8qOR38XlurbP4Y,3403
 autogluon/core/problem_type.py,sha256=XJmMgeNBgS7u43pDK-spTivatPyh_INOXveEXwQt-Rw,2993
-autogluon/core/version.py,sha256=AGq81lVpmmLWoZzbwCzgbSN_eMpjRM7ZUk1rDC3lkhc,90
+autogluon/core/version.py,sha256=TigPRfmNcZyeXSLH943LINQSmE-Pv8VStKn7i2xjUOM,90
 autogluon/core/augmentation/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autogluon/core/augmentation/distill_utils.py,sha256=JBlp2WOMNKoJv8aKVwJVRQSalSk8jx36HM7-k_VvkhY,9404
 autogluon/core/calibrate/__init__.py,sha256=eU6qLj7DKUhaz2HHNHDrfroRaLM-mhuSncK_v1UP4F8,62
@@ -39,7 +39,7 @@ autogluon/core/metrics/softclass_metrics.py,sha256=inn35DfftLZey0mK3OuMJPzm58eZg
 autogluon/core/models/__init__.py,sha256=dg3onYq5wW3-sfdNurnSIGpX0rpEjG_abgzyfwDM77M,408
 autogluon/core/models/_utils.py,sha256=qswE9n1ge1AJSExgstEbrZiMFmMRa4Mf5Sz8D9-XU6c,2091
 autogluon/core/models/abstract/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-autogluon/core/models/abstract/_tags.py,sha256=9GZMHdbek9D5-8X-vtCXNw-eICFwoPZ82zn3j0iYyo0,2962
+autogluon/core/models/abstract/_tags.py,sha256=Qr_3an0ZMig24S3OwISa-nTFfWHQe3pwPTiXq4zlEec,3409
 autogluon/core/models/abstract/abstract_model.py,sha256=bi0WOb51WJG6tAWuo3GKKgAfSSrBStwKYHEkDhVQeD4,120425
 autogluon/core/models/abstract/abstract_nn_model.py,sha256=IId0ivO8uVvmpnK9OiM2CtPVrP1ewOaQQKtQUDtK7_k,4818
 autogluon/core/models/abstract/model_trial.py,sha256=PKEo1jfLSBCOLM42QE5VBD1u41MaVMRk31zhNhLiqTw,5035
@@ -47,7 +47,7 @@ autogluon/core/models/dummy/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMp
 autogluon/core/models/dummy/_dummy_quantile_regressor.py,sha256=i-ZW2flJ60jsMfMK24IP39Xwc55-UlBDvHmqanIf29Q,664
 autogluon/core/models/dummy/dummy_model.py,sha256=at2FZSM2_LuAQ78E2YrRCRt3UaKMyyOnc6p2rtZgA2w,1414
 autogluon/core/models/ensemble/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-autogluon/core/models/ensemble/bagged_ensemble_model.py,sha256=OImzuazZL_5J4tXz4tt-vdUYDvAClA2uIRyY2OLdpjQ,69855
+autogluon/core/models/ensemble/bagged_ensemble_model.py,sha256=HuyRqdtsdN2z_t9Fa9qWN3U5dz3O7MGYn2qe4BKU9Go,71600
 autogluon/core/models/ensemble/fold_fitting_strategy.py,sha256=01vzNVvE4FIFgD6YqbhK63XoUlSztnVFsrDdsoqm75U,47021
 autogluon/core/models/ensemble/ray_parallel_fold_fitting_strategy.py,sha256=8RASa-eV6n9kUgbqQHNt7k4IrvuB9NdrunIMLYOLwgA,2068
 autogluon/core/models/ensemble/stacker_ensemble_model.py,sha256=DuDXgozvG9JYYkRvGACA7EXDAtj3Tz_uAjXTfxu5tFg,18041
@@ -89,11 +89,11 @@ autogluon/core/utils/utils.py,sha256=K05ewQuGauLnVaYwccNDk1moUDg2EEzdSlq8gsw6JVM
 autogluon/core/utils/version_utils.py,sha256=5-r8hLRKTaZbj5qo2uzE_2E4casH49Ye3WyeHlgHuz4,3252
 autogluon/core/utils/loaders/__init__.py,sha256=W5FAdQvpDcn_uisqJrlSAObWVta-YjJLKGN3NCbEgIo,109
 autogluon/core/utils/savers/__init__.py,sha256=bGWciSxAkj6u06vOC4pTvr22f_1ey0glgvmjCMEOm78,89
-autogluon.core-1.2.1b20250110.dist-info/LICENSE,sha256=CeipvOyAZxBGUsFoaFqwkx54aPnIKEtm9a5u2uXxEws,10142
-autogluon.core-1.2.1b20250110.dist-info/METADATA,sha256=umPV72cojnwuMAjsfFTdvlHikJWd3oKznl6b7O27TVY,12328
-autogluon.core-1.2.1b20250110.dist-info/NOTICE,sha256=7nPQuj8Kp-uXsU0S5so3-2dNU5EctS5hDXvvzzehd7E,114
-autogluon.core-1.2.1b20250110.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
-autogluon.core-1.2.1b20250110.dist-info/namespace_packages.txt,sha256=giERA4R78OkJf2ijn5slgjURlhRPzfLr7waIcGkzYAo,10
-autogluon.core-1.2.1b20250110.dist-info/top_level.txt,sha256=giERA4R78OkJf2ijn5slgjURlhRPzfLr7waIcGkzYAo,10
-autogluon.core-1.2.1b20250110.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
-autogluon.core-1.2.1b20250110.dist-info/RECORD,,
+autogluon.core-1.2.1b20250111.dist-info/LICENSE,sha256=CeipvOyAZxBGUsFoaFqwkx54aPnIKEtm9a5u2uXxEws,10142
+autogluon.core-1.2.1b20250111.dist-info/METADATA,sha256=SjvIvISrTQooEHQlUsjRb26e1L0lt00JCd1_tZ9JLgE,12328
+autogluon.core-1.2.1b20250111.dist-info/NOTICE,sha256=7nPQuj8Kp-uXsU0S5so3-2dNU5EctS5hDXvvzzehd7E,114
+autogluon.core-1.2.1b20250111.dist-info/WHEEL,sha256=tZoeGjtWxWRfdplE7E3d45VPlLNQnvbKiYnx7gwAy8A,92
+autogluon.core-1.2.1b20250111.dist-info/namespace_packages.txt,sha256=giERA4R78OkJf2ijn5slgjURlhRPzfLr7waIcGkzYAo,10
+autogluon.core-1.2.1b20250111.dist-info/top_level.txt,sha256=giERA4R78OkJf2ijn5slgjURlhRPzfLr7waIcGkzYAo,10
+autogluon.core-1.2.1b20250111.dist-info/zip-safe,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
+autogluon.core-1.2.1b20250111.dist-info/RECORD,,