autogluon.timeseries 1.1.0b20240416__tar.gz → 1.1.1__tar.gz

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.timeseries
-Version: 1.1.0b20240416
+Version: 1.1.1
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community
@@ -107,7 +107,7 @@ Description:
         This library is licensed under the Apache 2.0 License.
         
 Platform: UNKNOWN
-Classifier: Development Status :: 4 - Beta
+Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Education
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/setup.py

@@ -30,7 +30,7 @@ install_requires = [
     "pytorch_lightning",  # version range defined in `core/_setup_utils.py`
     "transformers[sentencepiece]",  # version range defined in `core/_setup_utils.py`
     "accelerate",  # version range defined in `core/_setup_utils.py`
-    "gluonts>=0.14.0,<0.14.4",  # 0.14.4 caps pandas<2.2
+    "gluonts==0.15.1",
     "networkx",  # version range defined in `core/_setup_utils.py`
     # TODO: update statsforecast to v1.5.0 - resolve antlr4-python3-runtime dependency clash with multimodal
     "statsforecast>=1.4.0,<1.5",

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/__init__.py

@@ -6,6 +6,7 @@ from .gluonts import (
     PatchTSTModel,
     SimpleFeedForwardModel,
     TemporalFusionTransformerModel,
+    TiDEModel,
     WaveNetModel,
 )
 from .local import (
@@ -55,6 +56,7 @@ __all__ = [
     "SimpleFeedForwardModel",
     "TemporalFusionTransformerModel",
     "ThetaModel",
+    "TiDEModel",
     "WaveNetModel",
     "ZeroModel",
 ]

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/autogluon_tabular/mlforecast.py

@@ -86,6 +86,28 @@ class AbstractMLForecastModel(AbstractTimeSeriesModel):
         self._residuals_std_per_item: Optional[pd.Series] = None
         self._avg_residuals_std: Optional[float] = None
         self._train_target_median: Optional[float] = None
+        self._non_boolean_real_covariates: List[str] = []
+
+    @property
+    def tabular_predictor_path(self) -> str:
+        return os.path.join(self.path, "tabular_predictor")
+
+    def save(self, path: str = None, verbose: bool = True) -> str:
+        assert "mean" in self._mlf.models_, "TabularPredictor must be trained before saving"
+        tabular_predictor = self._mlf.models_["mean"].predictor
+        self._mlf.models_["mean"].predictor = None
+        save_path = super().save(path=path, verbose=verbose)
+        self._mlf.models_["mean"].predictor = tabular_predictor
+        return save_path
+
+    @classmethod
+    def load(
+        cls, path: str, reset_paths: bool = True, load_oof: bool = False, verbose: bool = True
+    ) -> "AbstractTimeSeriesModel":
+        model = super().load(path=path, reset_paths=reset_paths, load_oof=load_oof, verbose=verbose)
+        assert "mean" in model._mlf.models_, "Loaded model doesn't have a trained TabularPredictor"
+        model._mlf.models_["mean"].predictor = TabularPredictor.load(model.tabular_predictor_path)
+        return model
 
     def preprocess(self, data: TimeSeriesDataFrame, is_train: bool = False, **kwargs) -> Any:
         if is_train:
@@ -252,10 +274,9 @@ class AbstractMLForecastModel(AbstractTimeSeriesModel):
         if static_features is not None:
             df = pd.merge(df, static_features, how="left", on=ITEMID, suffixes=(None, "_static_feat"))
 
-        for col in self.metadata.known_covariates_real:
+        for col in self._non_boolean_real_covariates:
             # Normalize non-boolean features using mean_abs scaling
-            if not df[col].isin([0, 1]).all():
-                df[f"__scaled_{col}"] = df[col] / df[col].abs().groupby(df[ITEMID]).mean().reindex(df[ITEMID]).values
+            df[f"__scaled_{col}"] = df[col] / df[col].abs().groupby(df[ITEMID]).mean().reindex(df[ITEMID]).values
 
         # Convert float64 to float32 to reduce memory usage
         float64_cols = list(df.select_dtypes(include="float64"))
@@ -277,6 +298,9 @@ class AbstractMLForecastModel(AbstractTimeSeriesModel):
         self._check_fit_params()
         fit_start_time = time.time()
         self._train_target_median = train_data[self.target].median()
+        for col in self.metadata.known_covariates_real:
+            if not train_data[col].isin([0, 1]).all():
+                self._non_boolean_real_covariates.append(col)
         # TabularEstimator is passed to MLForecast later to include tuning_data
         model_params = self._get_model_params()
 
@@ -292,7 +316,7 @@ class AbstractMLForecastModel(AbstractTimeSeriesModel):
 
         estimator = TabularEstimator(
             predictor_init_kwargs={
-                "path": os.path.join(self.path, "tabular_predictor"),
+                "path": self.tabular_predictor_path,
                 "verbosity": verbosity - 2,
                 "label": MLF_TARGET,
                 **self._get_extra_tabular_init_kwargs(),
@@ -349,7 +373,12 @@ class AbstractMLForecastModel(AbstractTimeSeriesModel):
                 "Fallback model SeasonalNaive is used for these time series."
             )
             data_short = data.query("item_id in @short_series")
-            seasonal_naive = SeasonalNaiveModel(freq=self.freq, prediction_length=self.prediction_length)
+            seasonal_naive = SeasonalNaiveModel(
+                freq=self.freq,
+                prediction_length=self.prediction_length,
+                target=self.target,
+                quantile_levels=self.quantile_levels,
+            )
             seasonal_naive.fit(train_data=data_short)
             forecast_for_short_series = seasonal_naive.predict(data_short)
 

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/chronos/model.py

@@ -16,27 +16,27 @@ logger = logging.getLogger(__name__)
 
 # allowed HuggingFace model paths with custom parameter definitions
 MODEL_CONFIGS = {
-    "amazon/chronos-t5-tiny": {
+    "chronos-t5-tiny": {
         "num_gpus": 0,  # minimum number of required GPUs
         "default_torch_dtype": "auto",
         "default_batch_size": 16,
     },
-    "amazon/chronos-t5-mini": {
+    "chronos-t5-mini": {
         "num_gpus": 0,
         "default_torch_dtype": "auto",
         "default_batch_size": 16,
     },
-    "amazon/chronos-t5-small": {
+    "chronos-t5-small": {
         "num_gpus": 1,
         "default_torch_dtype": "bfloat16",
         "default_batch_size": 16,
     },
-    "amazon/chronos-t5-base": {
+    "chronos-t5-base": {
         "num_gpus": 1,
         "default_torch_dtype": "bfloat16",
         "default_batch_size": 16,
     },
-    "amazon/chronos-t5-large": {
+    "chronos-t5-large": {
         "num_gpus": 1,
         "default_torch_dtype": "bfloat16",
         "default_batch_size": 8,
@@ -45,11 +45,11 @@ MODEL_CONFIGS = {
 
 
 MODEL_ALIASES = {
-    "tiny": "amazon/chronos-t5-tiny",
-    "mini": "amazon/chronos-t5-mini",
-    "small": "amazon/chronos-t5-small",
-    "base": "amazon/chronos-t5-base",
-    "large": "amazon/chronos-t5-large",
+    "tiny": "autogluon/chronos-t5-tiny",
+    "mini": "autogluon/chronos-t5-mini",
+    "small": "autogluon/chronos-t5-small",
+    "base": "autogluon/chronos-t5-base",
+    "large": "autogluon/chronos-t5-large",
 }
 
 
@@ -75,10 +75,10 @@ class ChronosModel(AbstractTimeSeriesModel):
 
     Other Parameters
     ----------------
-    model_path: str, default = "amazon/chronos-t5-small"
+    model_path: str, default = "autogluon/chronos-t5-small"
         Model path used for the model, i.e., a HuggingFace transformers ``name_or_path``. Can be a
         compatible model name on HuggingFace Hub or a local path to a model directory. Original
-        Chronos models (i.e., ``amazon/chronos-t5-{model_size}``) can be specified with aliases
+        Chronos models (i.e., ``autogluon/chronos-t5-{model_size}``) can be specified with aliases
         ``tiny``, ``mini`` , ``small``, ``base``, and ``large``.
     batch_size : int, default = 16
         Size of batches used during inference
@@ -106,7 +106,7 @@ class ChronosModel(AbstractTimeSeriesModel):
 
     # default number of samples for prediction
     default_num_samples: int = 20
-    default_model_path = "amazon/chronos-t5-small"
+    default_model_path = "autogluon/chronos-t5-small"
     maximum_context_length = 512
 
     def __init__(
@@ -185,9 +185,10 @@ class ChronosModel(AbstractTimeSeriesModel):
     @property
     def ag_default_config(self) -> Dict[str, Any]:
         """The default configuration of the model used by AutoGluon if the model is one of those
-        defined in MODEL_CONFIGS. For now, these are ``amazon/chronos-t5-*`` family of models.
+        defined in MODEL_CONFIGS. For now, these are ``autogluon/chronos-t5-*`` family of models.
         """
-        return MODEL_CONFIGS.get(self.model_path, {})
+        model_name = str(self.model_path).split("/")[-1]
+        return MODEL_CONFIGS.get(model_name, {})
 
     @property
     def min_num_gpus(self) -> int:

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/chronos/pipeline.py

@@ -163,7 +163,7 @@ class MeanScaleUniformBins(ChronosTokenizer):
     def output_transform(self, samples: torch.Tensor, scale: torch.Tensor) -> torch.Tensor:
         scale_unsqueezed = scale.unsqueeze(-1).unsqueeze(-1)
         indices = torch.clamp(
-            samples - self.config.n_special_tokens,
+            samples - self.config.n_special_tokens - 1,
             min=0,
             max=len(self.centers) - 1,
         )

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/gluonts/__init__.py

@@ -4,6 +4,7 @@ from .torch.models import (
     PatchTSTModel,
     SimpleFeedForwardModel,
     TemporalFusionTransformerModel,
+    TiDEModel,
     WaveNetModel,
 )
 
@@ -13,5 +14,6 @@ __all__ = [
     "PatchTSTModel",
     "SimpleFeedForwardModel",
     "TemporalFusionTransformerModel",
+    "TiDEModel",
     "WaveNetModel",
 ]

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/gluonts/torch/models.py

@@ -424,3 +424,94 @@ class WaveNetModel(AbstractGluonTSModel):
         init_kwargs.setdefault("time_features", get_time_features_for_frequency(self.freq))
         init_kwargs.setdefault("num_parallel_samples", self.default_num_samples)
         return init_kwargs
+
+
+class TiDEModel(AbstractGluonTSModel):
+    """Time series dense encoder model from [Das2023]_.
+
+    Based on `gluonts.torch.model.tide.TiDEEstimator <https://ts.gluon.ai/stable/api/gluonts/gluonts.torch.model.tide.html>`_.
+    See GluonTS documentation for additional hyperparameters.
+
+
+    References
+    ----------
+    .. [Das2023] Das, Abhimanyu, et al.
+        "Long-term Forecasting with TiDE: Time-series Dense Encoder."
+        Transactions of Machine Learning Research. 2023.
+
+    Other Parameters
+    ----------------
+    context_length : int, default = max(64, 2 * prediction_length)
+        Number of past values used for prediction.
+    disable_static_features : bool, default = False
+        If True, static features won't be used by the model even if they are present in the dataset.
+        If False, static features will be used by the model if they are present in the dataset.
+    disable_known_covariates : bool, default = False
+        If True, known covariates won't be used by the model even if they are present in the dataset.
+        If False, known covariates will be used by the model if they are present in the dataset.
+    disable_past_covariates : bool, default = False
+        If True, past covariates won't be used by the model even if they are present in the dataset.
+        If False, past covariates will be used by the model if they are present in the dataset.
+    feat_proj_hidden_dim : int, default = 4
+        Size of the feature projection layer.
+    encoder_hidden_dim : int, default = 4
+        Size of the dense encoder layer.
+    decoder_hidden_dim : int, default = 4
+        Size of the dense decoder layer.
+    temporal_hidden_dim : int, default = 4
+        Size of the temporal decoder layer.
+    distr_hidden_dim : int, default = 4
+        Size of the distribution projection layer.
+    num_layers_encoder : int, default = 1
+        Number of layers in dense encoder.
+    num_layers_decoder : int, default = 1
+        Number of layers in dense decoder.
+    decoder_output_dim : int, default = 4
+        Output size of the dense decoder.
+    dropout_rate : float, default = 0.3
+        Dropout regularization parameter.
+    num_feat_dynamic_proj : int, default = 2
+        Output size of feature projection layer.
+    embedding_dimension : int, default = [16] * num_feat_static_cat
+        Dimension of the embeddings for categorical features
+    layer_norm : bool, default = False
+        Should layer normalization be enabled?
+    scaling : {"mean", "std", None}, default = "mean"
+        Scaling applied to the inputs. One of ``"mean"`` (mean absolute scaling), ``"std"`` (standardization), ``None`` (no scaling).
+    max_epochs : int, default = 100
+        Number of epochs the model will be trained for
+    batch_size : int, default = 64
+        Size of batches used during training
+    predict_batch_size : int, default = 500
+        Size of batches used during prediction.
+    num_batches_per_epoch : int, default = 50
+        Number of batches processed every epoch
+    lr : float, default = 1e-3,
+        Learning rate used during training
+    trainer_kwargs : dict, optional
+        Optional keyword arguments passed to ``lightning.Trainer``.
+    early_stopping_patience : int or None, default = 20
+        Early stop training if the validation loss doesn't improve for this many epochs.
+    keep_lightning_logs : bool, default = False
+        If True, ``lightning_logs`` directory will NOT be removed after the model finished training.
+    """
+
+    supports_known_covariates = True
+    supports_static_features = True
+
+    @property
+    def default_context_length(self) -> int:
+        return min(512, max(64, 2 * self.prediction_length))
+
+    def _get_estimator_class(self) -> Type[GluonTSEstimator]:
+        from gluonts.torch.model.tide import TiDEEstimator
+
+        return TiDEEstimator
+
+    def _get_estimator_init_args(self) -> Dict[str, Any]:
+        init_kwargs = super()._get_estimator_init_args()
+        init_kwargs["num_feat_static_cat"] = self.num_feat_static_cat
+        init_kwargs["num_feat_static_real"] = self.num_feat_static_real
+        init_kwargs["cardinality"] = self.feat_static_cat_cardinality
+        init_kwargs["num_feat_dynamic_real"] = self.num_feat_dynamic_real
+        return init_kwargs

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/models/presets.py

@@ -31,6 +31,7 @@ from . import (
     SimpleFeedForwardModel,
     TemporalFusionTransformerModel,
     ThetaModel,
+    TiDEModel,
     WaveNetModel,
     ZeroModel,
 )
@@ -51,6 +52,7 @@ MODEL_TYPES = dict(
     DLinear=DLinearModel,
     PatchTST=PatchTSTModel,
     TemporalFusionTransformer=TemporalFusionTransformerModel,
+    TiDE=TiDEModel,
     WaveNet=WaveNetModel,
     RecursiveTabular=RecursiveTabularModel,
     DirectTabular=DirectTabularModel,
@@ -93,6 +95,7 @@ DEFAULT_MODEL_PRIORITY = dict(
     # Models that can early stop are trained at the end
     TemporalFusionTransformer=45,
     DeepAR=40,
+    TiDE=30,
     PatchTST=30,
     # Models below are not included in any presets
     WaveNet=25,

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/predictor.py

@@ -146,7 +146,7 @@ class TimeSeriesPredictor(TimeSeriesPredictorDeprecatedMixin):
     """
 
     predictor_file_name = "predictor.pkl"
-    _predictor_version_file_name = "__version__"
+    _predictor_version_file_name = "version.txt"
     _predictor_log_file_name = "predictor_log.txt"
 
     def __init__(
@@ -190,7 +190,7 @@ class TimeSeriesPredictor(TimeSeriesPredictorDeprecatedMixin):
             raise ValueError(f"Target column {self.target} cannot be one of the known covariates.")
         self.known_covariates_names = list(known_covariates_names)
 
-        self.prediction_length = prediction_length
+        self.prediction_length = int(prediction_length)
         # For each validation fold, all time series in training set must have length >= _min_train_length
         self._min_train_length = max(self.prediction_length + 1, 5)
         self.freq = freq
@@ -580,7 +580,7 @@ class TimeSeriesPredictor(TimeSeriesPredictorDeprecatedMixin):
 
             * "num_trials": How many HPO trials to run
             * "scheduler": Which scheduler to use. Valid values:
-                * "local": Local shceduler that schedules trials FIFO
+                * "local": Local scheduler that schedules trials FIFO
             * "searcher": Which searching algorithm to use. Valid values:
                 * "local_random": Uses the "random" searcher
                 * "random": Perform random search
@@ -1041,14 +1041,40 @@ class TimeSeriesPredictor(TimeSeriesPredictorDeprecatedMixin):
 
     @classmethod
     def _load_version_file(cls, path: str) -> str:
+        """
+        Loads the version file that is part of the saved predictor artifact.
+
+        Parameters
+        ----------
+        path: str
+            The path that would be used to load the predictor via `predictor.load(path)`
+
+        Returns
+        -------
+        The version of AutoGluon used to fit the predictor, as a string.
+
+        """
         version_file_path = os.path.join(path, cls._predictor_version_file_name)
-        version = load_str.load(path=version_file_path)
+        try:
+            version = load_str.load(path=version_file_path)
+        except:
+            # Loads the old version file used in `autogluon.timeseries<=1.1.0`, named `__version__`.
+            # This file name was changed because Kaggle does not allow uploading files named `__version__`.
+            version_file_path = os.path.join(path, "__version__")
+            version = load_str.load(path=version_file_path)
         return version
 
     @classmethod
     def load(cls, path: Union[str, Path], require_version_match: bool = True) -> "TimeSeriesPredictor":
         """Load an existing ``TimeSeriesPredictor`` from given ``path``.
 
+        .. warning::
+
+            :meth:`autogluon.timeseries.TimeSeriesPredictor.load` uses `pickle` module implicitly, which is known to
+            be insecure. It is possible to construct malicious pickle data which will execute arbitrary code during
+            unpickling. Never load data that could have come from an untrusted source, or that could have been tampered
+            with. **Only load data you trust.**
+
         Parameters
         ----------
         path : str or pathlib.Path
@@ -1077,7 +1103,7 @@ class TimeSeriesPredictor(TimeSeriesPredictorDeprecatedMixin):
         except:
             logger.warning(
                 f'WARNING: Could not find version file at "{os.path.join(path, cls._predictor_version_file_name)}".\n'
-                f"This means that the predictor was fit in a version `<=0.7.0`."
+                f"This means that the predictor was fit in an AutoGluon version `<=0.7.0`."
             )
             version_saved = "Unknown (Likely <=0.7.0)"
 

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/trainer/abstract_trainer.py

@@ -292,6 +292,10 @@ class AbstractTimeSeriesTrainer(SimpleAbstractTrainer):
         self.cache_predictions = cache_predictions
         self.hpo_results = {}
 
+        if self._cached_predictions_path.exists():
+            logger.debug(f"Removing existing cached predictions file {self._cached_predictions_path}")
+            self._cached_predictions_path.unlink()
+
     def save_train_data(self, data: TimeSeriesDataFrame, verbose: bool = True) -> None:
         path = os.path.join(self.path_data, "train.pkl")
         save_pkl.save(path=path, object=data, verbose=verbose)

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon/timeseries/version.py

@@ -1,3 +1,3 @@
 """This is the autogluon version file."""
-__version__ = '1.1.0b20240416'
+__version__ = '1.1.1'
 __lite__ = False

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon.timeseries.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.timeseries
-Version: 1.1.0b20240416
+Version: 1.1.1
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community
@@ -107,7 +107,7 @@ Description:
         This library is licensed under the Apache 2.0 License.
         
 Platform: UNKNOWN
-Classifier: Development Status :: 4 - Beta
+Classifier: Development Status :: 5 - Production/Stable
 Classifier: Intended Audience :: Education
 Classifier: Intended Audience :: Developers
 Classifier: Intended Audience :: Science/Research

{autogluon.timeseries-1.1.0b20240416 → autogluon.timeseries-1.1.1}/src/autogluon.timeseries.egg-info/requires.txt

@@ -2,12 +2,12 @@ joblib<2,>=1.1
 numpy<1.29,>=1.21
 scipy<1.13,>=1.5.4
 pandas<2.3.0,>=2.0.0
-torch<2.2,>=2.1
-lightning<2.2,>=2.1
-pytorch_lightning<2.2,>=2.1
-transformers[sentencepiece]<4.39.0,>=4.38.0
+torch<2.4,>=2.2
+lightning<2.4,>=2.2
+pytorch_lightning<2.4,>=2.2
+transformers[sentencepiece]<4.41.0,>=4.38.0
 accelerate<0.22.0,>=0.21.0
-gluonts<0.14.4,>=0.14.0
+gluonts==0.15.1
 networkx<4,>=3.0
 statsforecast<1.5,>=1.4.0
 mlforecast<0.10.1,>=0.10.0
@@ -15,9 +15,9 @@ utilsforecast<0.0.11,>=0.0.10
 tqdm<5,>=4.38
 orjson~=3.9
 tensorboard<3,>=2.9
-autogluon.core[raytune]==1.1.0b20240416
-autogluon.common==1.1.0b20240416
-autogluon.tabular[catboost,lightgbm,xgboost]==1.1.0b20240416
+autogluon.core[raytune]==1.1.1
+autogluon.common==1.1.1
+autogluon.tabular[catboost,lightgbm,xgboost]==1.1.1
 
 [all]
 optimum[onnxruntime]<1.19,>=1.17