autogluon.core 1.2.1b20250109__tar.gz → 1.2.1b20250110__tar.gz

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.core
-Version: 1.2.1b20250109
+Version: 1.2.1b20250110
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community
@@ -21,7 +21,7 @@ Description:
         [![Python Versions](https://img.shields.io/badge/python-3.9%20%7C%203.10%20%7C%203.11%20%7C%203.12-blue)](https://pypi.org/project/autogluon/)
         [![Downloads](https://pepy.tech/badge/autogluon/month)](https://pepy.tech/project/autogluon)
         [![GitHub license](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](./LICENSE)
-        [![Discord](https://img.shields.io/discord/1043248669505368144?logo=discord&style=flat)](https://discord.gg/wjUmjqAc2N)
+        [![Discord](https://img.shields.io/discord/1043248669505368144?color=7289da&label=Discord&logo=discord&logoColor=ffffff)](https://discord.gg/wjUmjqAc2N)
         [![Twitter](https://img.shields.io/twitter/follow/autogluon?style=social)](https://twitter.com/autogluon)
         [![Continuous Integration](https://github.com/autogluon/autogluon/actions/workflows/continuous_integration.yml/badge.svg)](https://github.com/autogluon/autogluon/actions/workflows/continuous_integration.yml)
         [![Platform Tests](https://github.com/autogluon/autogluon/actions/workflows/platform_tests-command.yml/badge.svg?event=schedule)](https://github.com/autogluon/autogluon/actions/workflows/platform_tests-command.yml)

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/setup.py

@@ -48,11 +48,11 @@ install_requires = (
 
 extras_require = {
     "ray": [
-        "ray[default]>=2.10.0,<2.40",  # sync with common/src/autogluon/common/utils/try_import.py
+        "ray[default]>=2.10.0,<2.41",  # sync with common/src/autogluon/common/utils/try_import.py
     ],
     "raytune": [
         "pyarrow>=15.0.0",  # cap Pyarrow to fix source installation - https://github.com/autogluon/autogluon/issues/4519
-        "ray[default,tune]>=2.10.0,<2.40",  # sync with common/src/autogluon/common/utils/try_import.py
+        "ray[default,tune]>=2.10.0,<2.41",  # sync with common/src/autogluon/common/utils/try_import.py
         # TODO: consider alternatives as hyperopt is not actively maintained.
         "hyperopt>=0.2.7,<0.2.8",  # This is needed for the bayes search to work.
         # 'GPy>=1.10.0,<1.11.0'  # TODO: Enable this once PBT/PB2 are supported by ray lightning

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon/core/metrics/__init__.py

@@ -47,9 +47,22 @@ class Scorer(object, metaclass=ABCMeta):
     kwargs : dict, optional
         kwargs to pass to score_func when called.
         For example, kwargs = {"beta": 2} when using sklearn.metrics.fbeta_score where beta is a required argument.
+    needs_pos_label : bool, default = False
+        If True, indicates that the metric requires a positive label specified via the `pos_label` argument.
+        Example metrics that require `pos_label`: ["f1", "precision", "recall"]
+        Currently this is used for unit testing purposes and does not impact the Scorer object.
     """
 
-    def __init__(self, name: str, score_func: callable, optimum: float, sign: int, kwargs: dict = None):
+    def __init__(
+        self,
+        name: str,
+        score_func: callable,
+        optimum: float,
+        sign: int,
+        kwargs: dict = None,
+        *,
+        needs_pos_label: bool = False,
+    ):
         self.name = name
         if kwargs is None:
             kwargs = dict()
@@ -59,6 +72,7 @@ class Scorer(object, metaclass=ABCMeta):
         if sign != 1 and sign != -1:
             raise ValueError(f"sign must be one of [1, -1], but was instead {sign}")
         self._sign = sign
+        self._needs_pos_label = needs_pos_label
         self.alias = set()
 
     def __call__(self, y_true, y_pred, sample_weight=None, **kwargs) -> float:
@@ -208,6 +222,15 @@ class Scorer(object, metaclass=ABCMeta):
         """If True, metric requires quantile predictions rather than predictions or prediction probabilities"""
         raise NotImplementedError
 
+    @property
+    def needs_pos_label(self) -> bool:
+        """
+        If True, metric requires pos_label to be specified. For most metrics, pos_label defaults to 1.
+        If unspecified and the user passes string values or values other than 0 and 1,
+        this can lead to exceptions or incorrect output.
+        """
+        return self._needs_pos_label
+
     score = __call__
 
 
@@ -419,6 +442,7 @@ def make_scorer(
     needs_class: bool = False,
     needs_threshold: bool = False,
     needs_quantile: bool = False,
+    needs_pos_label: bool = False,
     metric_kwargs: dict = None,
     **kwargs,
 ) -> Scorer:
@@ -474,6 +498,12 @@ def make_scorer(
         This only works for quantile regression.
         Examples: ["pinball_loss"]
 
+    needs_pos_label : bool, default=False
+        Whether score_func supports a pos_label argument.
+        For binary classification, input y_true and y_pred must contain the pos_label in order for the metric to be correctly calculated.
+        This only works for binary classification.
+        Examples: ["f1", "precision", "recall"]
+
     metric_kwargs : dict
         Additional parameters to be passed to score_func, merged with kwargs if both are present.
         metric_kwargs keys will override kwargs keys if keys are shared between them.
@@ -520,6 +550,7 @@ def make_scorer(
         optimum=optimum,
         sign=sign,
         kwargs=kwargs,
+        needs_pos_label=needs_pos_label,
     )
 
     if isinstance(needs_pred, bool) and needs_pred != scorer.needs_pred:
@@ -590,15 +621,17 @@ accuracy = make_scorer("accuracy", sklearn.metrics.accuracy_score, needs_class=T
 accuracy.add_alias("acc")
 
 balanced_accuracy = make_scorer("balanced_accuracy", classification_metrics.balanced_accuracy, needs_class=True)
-f1 = make_scorer("f1", sklearn.metrics.f1_score, needs_class=True)
 mcc = make_scorer("mcc", sklearn.metrics.matthews_corrcoef, needs_class=True)
 
 # Score functions that need decision values
 roc_auc = make_scorer("roc_auc", classification_metrics.customized_binary_roc_auc_score, greater_is_better=True, needs_threshold=True)
 
 average_precision = make_scorer("average_precision", sklearn.metrics.average_precision_score, needs_threshold=True)
-precision = make_scorer("precision", sklearn.metrics.precision_score, needs_class=True)
-recall = make_scorer("recall", sklearn.metrics.recall_score, needs_class=True)
+
+# Score functions that need pos_label
+f1 = make_scorer("f1", sklearn.metrics.f1_score, needs_class=True, needs_pos_label=True)
+precision = make_scorer("precision", sklearn.metrics.precision_score, needs_class=True, needs_pos_label=True)
+recall = make_scorer("recall", sklearn.metrics.recall_score, needs_class=True, needs_pos_label=True)
 
 # Register other metrics
 quadratic_kappa = make_scorer("quadratic_kappa", classification_metrics.quadratic_kappa, needs_class=True)
@@ -692,7 +725,6 @@ for scorer in [
 
 
 for _name, _metric in [("precision", sklearn.metrics.precision_score), ("recall", sklearn.metrics.recall_score), ("f1", sklearn.metrics.f1_score)]:
-    globals()[_name] = make_scorer(_name, _metric, needs_class=True)
     _add_scorer_to_metric_dict(metric_dict=BINARY_METRICS, scorer=globals()[_name])
     for average in ["macro", "micro", "weighted"]:
         qualified_name = "{0}_{1}".format(_name, average)
@@ -720,7 +752,7 @@ for _name, _metric, _kwargs in [
         _add_scorer_to_metric_dict(metric_dict=MULTICLASS_METRICS, scorer=globals()[qualified_name])
 
 
-METRICS = {
+METRICS: dict[str, dict[str, Scorer]] = {
     BINARY: BINARY_METRICS,
     MULTICLASS: MULTICLASS_METRICS,
     REGRESSION: REGRESSION_METRICS,
@@ -736,14 +768,16 @@ def _get_valid_metric_problem_types(metric: str):
     return problem_types_valid
 
 
-def get_metric(metric, problem_type=None, metric_type=None) -> Scorer:
+def get_metric(metric, problem_type: str = None, metric_type: str = None) -> Scorer:
     """Returns metric function by using its name if the metric is str.
     Performs basic check for metric compatibility with given problem type."""
+    if metric_type is None:
+        metric_type = "metric"
 
     if metric is not None and isinstance(metric, str):
         if metric == "soft_log_loss":
             if problem_type == QUANTILE:
-                raise ValueError(f"{metric_type}={metric} can not be used for quantile problems")
+                raise ValueError(f"{metric_type}='{metric}' can not be used for quantile problems")
             from .softclass_metrics import soft_log_loss
 
             return soft_log_loss

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon/core/metrics/classification_metrics.py

@@ -25,7 +25,13 @@ def balanced_accuracy(solution, prediction):
 
     if y_type == "binary":
         # Do not transform into any multiclass representation
-        pass
+        unique_sol = np.unique(solution)
+        unique_pred = np.unique(prediction)
+        classes = np.unique(np.concatenate((unique_sol, unique_pred)))
+        if set(classes) != {0, 1}:
+            pos_class = classes[-1]
+            solution = np.array([1 if i == pos_class else 0 for i in solution])
+            prediction = np.array([1 if i == pos_class else 0 for i in prediction])
 
     elif y_type == "multiclass":
         n = len(solution)

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon/core/models/abstract/abstract_model.py

@@ -1793,6 +1793,7 @@ class AbstractModel:
         fit_kwargs["num_classes"] = self.num_classes
         fit_kwargs["sample_weight"] = kwargs.get("sample_weight", None)
         fit_kwargs["sample_weight_val"] = kwargs.get("sample_weight_val", None)
+        fit_kwargs["verbosity"] = kwargs.get("verbosity", 2)
         train_fn_kwargs = dict(
             model_cls=model_cls,
             init_params=init_params,

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon/core/models/ensemble/bagged_ensemble_model.py

@@ -1428,6 +1428,7 @@ class BaggedEnsembleModel(AbstractModel):
         fit_kwargs["num_classes"] = self.num_classes
         fit_kwargs["sample_weight"] = kwargs.get("sample_weight", None)
         fit_kwargs["sample_weight_val"] = kwargs.get("sample_weight_val", None)
+        fit_kwargs["verbosity"] = kwargs.get("verbosity", 2)
         fit_kwargs.pop("time_limit", None)  # time_limit already set in hpo_executor
         train_fn_kwargs = dict(
             model_cls=model_cls,

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon/core/version.py

@@ -1,3 +1,3 @@
 """This is the autogluon version file."""
-__version__ = '1.2.1b20250109'
+__version__ = '1.2.1b20250110'
 __lite__ = False

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon.core.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.core
-Version: 1.2.1b20250109
+Version: 1.2.1b20250110
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community
@@ -21,7 +21,7 @@ Description:
         [![Python Versions](https://img.shields.io/badge/python-3.9%20%7C%203.10%20%7C%203.11%20%7C%203.12-blue)](https://pypi.org/project/autogluon/)
         [![Downloads](https://pepy.tech/badge/autogluon/month)](https://pepy.tech/project/autogluon)
         [![GitHub license](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](./LICENSE)
-        [![Discord](https://img.shields.io/discord/1043248669505368144?logo=discord&style=flat)](https://discord.gg/wjUmjqAc2N)
+        [![Discord](https://img.shields.io/discord/1043248669505368144?color=7289da&label=Discord&logo=discord&logoColor=ffffff)](https://discord.gg/wjUmjqAc2N)
         [![Twitter](https://img.shields.io/twitter/follow/autogluon?style=social)](https://twitter.com/autogluon)
         [![Continuous Integration](https://github.com/autogluon/autogluon/actions/workflows/continuous_integration.yml/badge.svg)](https://github.com/autogluon/autogluon/actions/workflows/continuous_integration.yml)
         [![Platform Tests](https://github.com/autogluon/autogluon/actions/workflows/platform_tests-command.yml/badge.svg?event=schedule)](https://github.com/autogluon/autogluon/actions/workflows/platform_tests-command.yml)

{autogluon.core-1.2.1b20250109 → autogluon.core-1.2.1b20250110}/src/autogluon.core.egg-info/requires.txt

@@ -7,25 +7,25 @@ tqdm<5,>=4.38
 requests
 matplotlib<3.11,>=3.7.0
 boto3<2,>=1.10
-autogluon.common==1.2.1b20250109
+autogluon.common==1.2.1b20250110
 
 [all]
-hyperopt<0.2.8,>=0.2.7
-ray[default,tune]<2.40,>=2.10.0
+ray[default,tune]<2.41,>=2.10.0
+ray[default]<2.41,>=2.10.0
 pyarrow>=15.0.0
-ray[default]<2.40,>=2.10.0
+hyperopt<0.2.8,>=0.2.7
 
 [ray]
-ray[default]<2.40,>=2.10.0
+ray[default]<2.41,>=2.10.0
 
 [raytune]
 pyarrow>=15.0.0
-ray[default,tune]<2.40,>=2.10.0
+ray[default,tune]<2.41,>=2.10.0
 hyperopt<0.2.8,>=0.2.7
 
 [tests]
-pytest
-flake8
 types-setuptools
 pytest-mypy
+flake8
+pytest
 types-requests