[MNT] remove soft dependency import warnings in modules and documente…

…d requirements to add these (#4398)

Previously, modules that contained objects with soft dependencies were raising import warnings.
This would raise a lot of warnings in crawling based utilities such as `all_estimators` (and cause those utilities to have a longer runtime than otherwise).

In many cases, the warnings would be raised without a clear reason visible to the user, and would confuse more than they would help - contributing to some user frustration.
The information is still raised as an error when users try to construct the estimators in question - and there cause and effect are much clearer. Therefore, I would argue that it is an improvement to get rid of the module warnings.

This PR removes both the existing warnings in estimator modules and the convention itself:

* removed all calls to `_check_soft_dependencies` with warning level at module import, in estimator modules
* removed the instruction to write such checks from all extension templates
* removed the bullet point in the dependency handling guide that asked to add these, and rewrote adjacent content for coherence

docs/source/developer_guide/dependencies.rst

@@ -46,12 +46,10 @@ To add an estimator with a soft dependency, ensure the following:
 *  the ``python_dependencies`` tag of the estimator is populated with a ``str``,
    or a ``list`` of ``str``, of import dependencies. Exceptions will automatically raised when constructing the estimator
    in an environment without the required packages.
-*  in the python module containing the estimator, the ``_check_soft_dependencies`` utility is called
-   at the top of the module, with ``severity="warning"``. This will raise an informative warning message already at module import.
-   See `here <https://github.com/sktime/sktime/blob/main/sktime/utils/validation/_dependencies.py>`__
 *  In a case where the package import differs from the package name, i.e., ``import package_string`` is different from
    ``pip install different-package-string`` (usually the case for packages containing a dash in the name), the ``_check_soft_dependencies``
-   utility should be used in ``__init__``. Both the warning and constructor call should use the ``package_import_alias`` argument for this.
+   utility should be called in ``__init__`` before the ``super.__init__`` call, with ``severity="error"``.
+   The ``package_import_alias`` argument should be used to pass the information on package and import strings.
 *  If the soft dependencies require specific python versions, the ``python_version``
    tag should also be populated, with a PEP 440 compliant version specification ``str`` such as ``"<3.10"`` or ``">3.6,~=3.8"``.
 *  If including docstring examples that use soft dependencies, ensure to skip doctest. To do this add a ``# doctest: +SKIP`` to the end of each

extension_templates/classification.py

@@ -42,11 +42,7 @@
 # todo: add any necessary imports here
 
 # todo: if any imports are sktime soft dependencies:
-#  * make sure to fill in the "python_dependencies" tag with the package import name
-#  * add a _check_soft_dependencies warning here, example:
-#
-# from sktime.utils.validation._dependencies import check_soft_dependencies
-# _check_soft_dependencies("soft_dependency_name", severity="warning")
+# make sure to fill in the "python_dependencies" tag with the package import name
 
 
 class MyTimeSeriesClassifier(BaseClassifier):

extension_templates/dist_kern_panel.py

@@ -30,11 +30,7 @@
 # todo: add any necessary imports here
 
 # todo: if any imports are sktime soft dependencies:
-#  * make sure to fill in the "python_dependencies" tag with the package import name
-#  * add a _check_soft_dependencies warning here, example:
-#
-# from sktime.utils.validation._dependencies import check_soft_dependencies
-# _check_soft_dependencies("soft_dependency_name", severity="warning")
+# make sure to fill in the "python_dependencies" tag with the package import name
 
 
 class MyTrafoPwPanel(BasePairwiseTransformerPanel):

extension_templates/dist_kern_tab.py

@@ -30,11 +30,7 @@
 # todo: add any necessary imports here
 
 # todo: if any imports are sktime soft dependencies:
-#  * make sure to fill in the "python_dependencies" tag with the package import name
-#  * add a _check_soft_dependencies warning here, example:
-#
-# from sktime.utils.validation._dependencies import check_soft_dependencies
-# _check_soft_dependencies("soft_dependency_name", severity="warning")
+# make sure to fill in the "python_dependencies" tag with the package import name
 
 
 class MyTrafoPw(BasePairwiseTransformer):

extension_templates/forecasting.py

@@ -51,11 +51,7 @@
 # todo: add any necessary imports here
 
 # todo: if any imports are sktime soft dependencies:
-#  * make sure to fill in the "python_dependencies" tag with the package import name
-#  * add a _check_soft_dependencies warning here, example:
-#
-# from sktime.utils.validation._dependencies import check_soft_dependencies
-# _check_soft_dependencies("soft_dependency_name", severity="warning")
+# make sure to fill in the "python_dependencies" tag with the package import name
 
 
 class MyForecaster(BaseForecaster):

extension_templates/param_est.py

@@ -38,11 +38,7 @@
 # todo: add any necessary imports here
 
 # todo: if any imports are sktime soft dependencies:
-#  * make sure to fill in the "python_dependencies" tag with the package import name
-#  * add a _check_soft_dependencies warning here, example:
-#
-# from sktime.utils.validation._dependencies import check_soft_dependencies
-# _check_soft_dependencies("soft_dependency_name", severity="warning")
+# make sure to fill in the "python_dependencies" tag with the package import name
 
 
 class MyTimeSeriesParamFitter(BaseParamFitter):

extension_templates/transformer.py

@@ -49,11 +49,7 @@
 # todo: add any necessary sktime internal imports here
 
 # todo: if any imports are sktime soft dependencies:
-#  * make sure to fill in the "python_dependencies" tag with the package import name
-#  * add a _check_soft_dependencies warning here, example:
-#
-# from sktime.utils.validation._dependencies import check_soft_dependencies
-# _check_soft_dependencies("soft_dependency_name", severity="warning")
+# make sure to fill in the "python_dependencies" tag with the package import name
 
 
 class MyTransformer(BaseTransformer):

sktime/alignment/dtw_python.py

@@ -12,14 +12,6 @@
 from sktime.alignment.base import BaseAligner
 from sktime.utils.validation._dependencies import _check_soft_dependencies
 
-_check_soft_dependencies(
-    "dtw-python",
-    package_import_alias={"dtw-python": "dtw"},
-    severity="warning",
-    obj="AlignerDTW or AlignerDTWfromDist",
-    suppress_import_stdout=True,
-)
-
 
 class AlignerDTW(BaseAligner):
     """Aligner interface for dtw-python.

sktime/annotation/adapters/_pyod.py

@@ -13,8 +13,6 @@
 
 import pandas as pd
 
-_check_soft_dependencies("pyod", severity="warning")
-
 
 class PyODAnnotator(BaseSeriesAnnotator):
     """Transformer that applies outlier detector from pyOD.

sktime/annotation/hmm_learn/gaussian.py

@@ -8,13 +8,10 @@
 """
 
 from sktime.annotation.hmm_learn import BaseHMMLearn
-from sktime.utils.validation._dependencies import _check_soft_dependencies
 
 __author__ = ["miraep8"]
 __all__ = ["GaussianHMM"]
 
-_check_soft_dependencies("hmmlearn.hmm", severity="warning")
-
 
 class GaussianHMM(BaseHMMLearn):
     """Hidden Markov Model with Gaussian emissions.

sktime/annotation/hmm_learn/gmm.py

@@ -8,15 +8,11 @@
 """
 
 from sktime.annotation.hmm_learn import BaseHMMLearn
-from sktime.utils.validation._dependencies import _check_soft_dependencies
 
 __author__ = ["miraep8"]
 __all__ = ["GMMHMM"]
 
 
-_check_soft_dependencies("hmmlearn.hmm", severity="warning")
-
-
 class GMMHMM(BaseHMMLearn):
     """
     Hidden Markov Model with Gaussian mixture emissions.

sktime/annotation/hmm_learn/poisson.py

@@ -8,13 +8,10 @@
 """
 
 from sktime.annotation.hmm_learn import BaseHMMLearn
-from sktime.utils.validation._dependencies import _check_soft_dependencies
 
 __author__ = ["klam-data", "pyyim", "mgorlin"]
 __all__ = ["PoissonHMM"]
 
-_check_soft_dependencies("hmmlearn.hmm", severity="warning")
-
 
 class PoissonHMM(BaseHMMLearn):
     """

sktime/classification/deep_learning/cnn.py

@@ -12,8 +12,6 @@
 from sktime.networks.cnn import CNNNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class CNNClassifier(BaseDeepClassifier):
     """Time Convolutional Neural Network (CNN), as described in [1]_.

sktime/classification/deep_learning/fcn.py

@@ -12,8 +12,6 @@
 from sktime.networks.fcn import FCNNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class FCNClassifier(BaseDeepClassifier):
     """Fully Connected Neural Network (FCN), as described in [1]_.

sktime/classification/deep_learning/lstmfcn.py

@@ -10,9 +10,6 @@
 
 from sktime.classification.deep_learning.base import BaseDeepClassifier
 from sktime.networks.lstmfcn import LSTMFCNNetwork
-from sktime.utils.validation._dependencies import _check_dl_dependencies
-
-_check_dl_dependencies(severity="warning")
 
 
 class LSTMFCNClassifier(BaseDeepClassifier):

sktime/classification/deep_learning/mlp.py

@@ -12,8 +12,6 @@
 from sktime.networks.mlp import MLPNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class MLPClassifier(BaseDeepClassifier):
     """Multi Layer Perceptron Network (MLP), as described in [1]_.

sktime/classification/deep_learning/resnet.py

@@ -12,8 +12,6 @@
 from sktime.networks.resnet import ResNetNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class ResNetClassifier(BaseDeepClassifier):
     """

sktime/classification/deep_learning/tapnet.py

@@ -18,8 +18,6 @@
 from sktime.networks.tapnet import TapNetNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class TapNetClassifier(BaseDeepClassifier):
     """Time series attentional prototype network (TapNet), as described in [1]_.

sktime/clustering/k_shapes.py

@@ -6,9 +6,6 @@
 from numpy.random import RandomState
 
 from sktime.clustering.base import BaseClusterer, TimeSeriesInstances
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("tslearn", severity="warning")
 
 
 class TimeSeriesKShapes(BaseClusterer):

sktime/clustering/kernel_k_means.py

@@ -6,9 +6,6 @@
 from numpy.random import RandomState
 
 from sktime.clustering.base import BaseClusterer, TimeSeriesInstances
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("tslearn", severity="warning")
 
 
 class TimeSeriesKernelKMeans(BaseClusterer):

sktime/forecasting/adapters/_hcrystalball.py

@@ -11,8 +11,6 @@
 from sktime.forecasting.base import BaseForecaster
 from sktime.utils.validation._dependencies import _check_soft_dependencies
 
-_check_soft_dependencies("hcrystalball", severity="warning")
-
 
 def _check_fh(fh, cutoff):
     if fh is not None:

sktime/forecasting/arima.py

@@ -7,9 +7,6 @@
 __all__ = ["AutoARIMA", "ARIMA"]
 
 from sktime.forecasting.base.adapters._pmdarima import _PmdArimaAdapter
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("pmdarima", severity="warning")
 
 
 class AutoARIMA(_PmdArimaAdapter):

sktime/forecasting/bats.py

@@ -13,9 +13,6 @@
 __all__ = ["BATS"]
 
 from sktime.forecasting.base.adapters import _TbatsAdapter
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("tbats", severity="warning")
 
 
 class BATS(_TbatsAdapter):

sktime/forecasting/fbprophet.py

@@ -9,9 +9,6 @@
 
 from sktime.forecasting.base._base import DEFAULT_ALPHA
 from sktime.forecasting.base.adapters import _ProphetAdapter
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("prophet", severity="warning")
 
 
 class Prophet(_ProphetAdapter):

sktime/forecasting/statsforecast.py

@@ -9,9 +9,6 @@
 from typing import Dict, Optional
 
 from sktime.forecasting.base.adapters._statsforecast import _StatsForecastAdapter
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("statsforecast", severity="warning")
 
 
 class StatsForecastAutoARIMA(_StatsForecastAdapter):

sktime/forecasting/tbats.py

@@ -13,9 +13,6 @@
 __all__ = ["TBATS"]
 
 from sktime.forecasting.base.adapters import _TbatsAdapter
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("tbats", severity="warning")
 
 
 class TBATS(_TbatsAdapter):

sktime/networks/cnn.py

@@ -6,8 +6,6 @@
 from sktime.networks.base import BaseDeepNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class CNNNetwork(BaseDeepNetwork):
     """Establish the network structure for a CNN.

sktime/networks/fcn.py

@@ -6,8 +6,6 @@
 from sktime.networks.base import BaseDeepNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class FCNNetwork(BaseDeepNetwork):
     """Establish the network structure for a FCN.

sktime/networks/lstmfcn.py

@@ -4,9 +4,6 @@
 __author__ = ["jnrusson1", "solen0id"]
 
 from sktime.networks.base import BaseDeepNetwork
-from sktime.utils.validation._dependencies import _check_dl_dependencies
-
-_check_dl_dependencies(severity="warning")
 
 
 class LSTMFCNNetwork(BaseDeepNetwork):

sktime/networks/lstmfcn_layers.py

@@ -2,11 +2,6 @@
 """Attention Layers used in by the LSTM-FCN Network. Ported over from sktime-dl."""
 
 
-from sktime.utils.validation._dependencies import _check_dl_dependencies
-
-_check_dl_dependencies(severity="warning")
-
-
 def make_attention_lstm():
     """Return AttentionLSTM class used by the LSTM-FCN Network."""
     from tensorflow.keras import activations

sktime/networks/mlp.py

@@ -6,8 +6,6 @@
 from sktime.networks.base import BaseDeepNetwork
 from sktime.utils.validation._dependencies import _check_dl_dependencies
 
-_check_dl_dependencies(severity="warning")
-
 
 class MLPNetwork(BaseDeepNetwork):
     """Establish the network structure for a MLP.

sktime/networks/tapnet.py

@@ -15,13 +15,6 @@
     _check_soft_dependencies,
 )
 
-_check_soft_dependencies(
-    "keras-self-attention",
-    package_import_alias={"keras-self-attention": "keras_self_attention"},
-    severity="warning",
-)
-_check_dl_dependencies(severity="warning")
-
 
 class TapNetNetwork(BaseDeepNetwork):
     """Establish Network structure for TapNet.

sktime/transformations/panel/catch22wrapper.py

@@ -14,9 +14,6 @@
 from sktime.transformations.base import BaseTransformer
 from sktime.transformations.panel import catch22
 from sktime.utils.validation import check_n_jobs
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("pycatch22", severity="warning")
 
 
 class Catch22Wrapper(BaseTransformer):

sktime/transformations/panel/signature_based/_compute.py

@@ -8,9 +8,6 @@
     _rescale_signature,
 )
 from sktime.transformations.panel.signature_based._window import _window_getter
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("esig", severity="warning")
 
 
 class _WindowSignatureTransform(BaseTransformer):

sktime/transformations/panel/signature_based/_rescaling.py

@@ -12,10 +12,6 @@
 
 import numpy as np
 
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("esig", severity="warning")
-
 
 def _rescale_path(path, depth):
     """Rescale input path by depth! ** (1 / depth).

sktime/transformations/panel/tsfresh.py

@@ -10,9 +10,6 @@
 from sktime.datatypes._panel._convert import from_nested_to_long
 from sktime.transformations.base import BaseTransformer
 from sktime.utils.validation import check_n_jobs
-from sktime.utils.validation._dependencies import _check_soft_dependencies
-
-_check_soft_dependencies("tsfresh", severity="warning")
 
 
 class _TSFreshFeatureExtractor(BaseTransformer):