[DOC] Fix some documentation warnings (#1581)

* chore: fix docs

- header Examples

- header Returns

- header References

* fix: documentation indentation for ChannelEnsembleClassifier

Author: Sebastian Schmidl

aeon/base/_meta.py

@@ -557,8 +557,8 @@ def _anytagis(self, tag_name, value, estimators):
         value : value of the tag to check for
         estimators : list of (str, estimator) pairs to query for the tag/value
 
-        Return
-        ------
+        Returns
+        -------
         bool : True iff at least one estimator in the list has value in tag tag_name
         """
         tagis = [est.get_tag(tag_name, value) == value for _, est in estimators]
@@ -591,8 +591,8 @@ def _anytag_notnone_val(self, tag_name, estimators):
         tag_name : str, name of the tag
         estimators : list of (str, estimator) pairs to query for the tag/value
 
-        Return
-        ------
+        Returns
+        -------
         tag_val : first non-'None' value of tag `tag_name` in estimator list.
         """
         for _, est in estimators:
@@ -711,8 +711,8 @@ def flatten(obj):
     list or tuple, tuple if obj was tuple, list otherwise
         flat iterable, containing non-list/tuple elements in obj in same order as in obj
 
-    Example
-    -------
+    Examples
+    --------
     >>> flatten([1, 2, [3, (4, 5)], 6])
     [1, 2, 3, 4, 5, 6]
     """
@@ -737,8 +737,8 @@ def unflatten(obj, template):
         has element bracketing exactly as `template`
             and elements in sequence exactly as `obj`
 
-    Example
-    -------
+    Examples
+    --------
     >>> unflatten([1, 2, 3, 4, 5, 6], [6, 3, [5, (2, 4)], 1])
     [1, 2, [3, (4, 5)], 6]
     """

aeon/benchmarking/results_loaders.py

@@ -189,8 +189,8 @@ def estimator_alias(name: str) -> str:
     name: str
         Standardized name as defined by NAME_ALIASES.
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.benchmarking.results_loaders import estimator_alias
     >>> estimator_alias("HIVECOTEV2")
     'HC2'
@@ -223,8 +223,8 @@ def get_available_estimators(task="classification", return_dataframe=True):
     pd.DataFrame or List
         Standardised name as defined by NAME_ALIASES.
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.benchmarking.results_loaders import get_available_estimators
     >>> cls = get_available_estimators("Classification")  # doctest: +SKIP
     """
@@ -307,8 +307,8 @@ def get_estimator_results(
         dataset names for keys and results as the value. If default only is an
         np.ndarray.
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.benchmarking.results_loaders import get_estimator_results
     >>> cls = ["HC2"]  # doctest: +SKIP
     >>> data = ["Chinatown", "Adiac"]  # doctest: +SKIP
@@ -376,8 +376,8 @@ def get_estimator_results_as_array(
         Each column is a results for a classifier, each row a dataset.
     if include_missing == false, returns names: an aligned list of names of included.
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.benchmarking.results_loaders import get_estimator_results
     >>> cls = ["HC2", "FreshPRINCE"] # doctest: +SKIP
     >>> data = ["Chinatown", "Adiac"] # doctest: +SKIP

aeon/classification/compose/_channel_ensemble.py

@@ -177,8 +177,7 @@ def _predict(self, X) -> np.ndarray:
 
 
 class ChannelEnsembleClassifier(_BaseChannelEnsembleClassifier):
-    """
-    Applies estimators to channels of an array.
+    """Applies estimators to channels of an array.
 
     This estimator allows different channels or channel subsets of the input
     to be transformed separately and the features generated by each
@@ -197,8 +196,8 @@ class ChannelEnsembleClassifier(_BaseChannelEnsembleClassifier):
             Estimator must support `fit` and `predict_proba`. Special-cased
             strings 'drop' and 'passthrough' are accepted as well, to
             indicate to drop the channels.
-        channels(s) : array-like of int, slice, boolean mask array. Integer channels
-        are indexed from 0
+        channels(s) : array-like of int, slice, boolean mask array
+            Integer channels are indexed from 0.
     remainder : {'drop', 'passthrough'} or estimator, default 'drop'
         By default, only the specified channels in `transformations` are
         transformed and combined in the output, and the non-specified
@@ -258,9 +257,9 @@ def get_test_params(cls, parameter_set="default"):
             Name of the set of test parameters to return, for use in tests. If no
             special parameters are defined for a value, will return `"default"` set.
             ChannelEnsembleClassifier provides the following special sets:
-                 "results_comparison" - used in some classifiers to compare against
-                    previously generated results where the default set of parameters
-                    cannot produce suitable probability estimates
+            - "results_comparison" - used in some classifiers to compare against
+              previously generated results where the default set of parameters
+              cannot produce suitable probability estimates
 
         Returns
         -------
@@ -293,8 +292,7 @@ def get_test_params(cls, parameter_set="default"):
 
 
 def _get_channel(X, key):
-    """
-    Get time series channel(s) from input data X.
+    """Get time series channel(s) from input data X.
 
     Supported input types (X): numpy arrays
 
@@ -311,7 +309,6 @@ def _get_channel(X, key):
         - only supported for dataframes
         - So no keys other than strings are allowed (while in principle you
           can use any hashable object as key).
-
     """
     # check whether we have string channel names or integers
     if _check_key_type(key, int):
@@ -340,8 +337,7 @@ def _get_channel(X, key):
 
 
 def _check_key_type(key, superclass):
-    """
-    Check that scalar, list or slice is of a certain type.
+    """Check that scalar, list or slice is of a certain type.
 
     This is only used in _get_channel and _get_channel_indices to check
     if the `key` (channel specification) is fully integer or fully string-like.
@@ -352,7 +348,6 @@ def _check_key_type(key, superclass):
         The channel specification to check
     superclass : int or str
         The type for which to check the `key`
-
     """
     if isinstance(key, superclass):
         return True
@@ -372,11 +367,9 @@ def _check_key_type(key, superclass):
 
 
 def _get_channel_indices(X, key):
-    """
-    Get feature channel indices for input data X and key.
+    """Get feature channel indices for input data X and key.
 
     For accepted values of `key`, see the docstring of _get_channel
-
     """
     n_channels = X.shape[1]
 

aeon/classification/shapelet_based/_rsast_classifier.py

@@ -38,8 +38,8 @@ class RSASTClassifier(BaseClassifier):
         Number of threads to use for the transform.
 
 
-    Reference
-    ---------
+    References
+    ----------
     .. [1] Varela, N. R., Mbouopda, M. F., & Nguifo, E. M. (2023). RSAST: Sampling
     Shapelets for Time Series Classification.
     https://hal.science/hal-04311309/
@@ -90,8 +90,8 @@ def _fit(self, X, y):
         y: array-like or list
             The class values for X.
 
-        Return
-        ------
+        Returns
+        -------
         self : RSASTClassifier
             This pipeline classifier
 
@@ -127,8 +127,8 @@ def _predict(self, X):
         X: np.ndarray shape (n_cases, n_channels, n_timepoints)
             The training input samples.
 
-        Return
-        ------
+        Returns
+        -------
         array-like or list
             Predicted class labels.
         """
@@ -142,8 +142,8 @@ def _predict_proba(self, X):
         X: np.ndarray shape (n_cases, n_channels, n_timepoints)
             The training input samples.
 
-        Return
-        ------
+        Returns
+        -------
         dists : np.ndarray shape (n_cases, n_timepoints)
             Predicted class probabilities.
         """

aeon/classification/shapelet_based/_sast_classifier.py

@@ -38,13 +38,13 @@ class SASTClassifier(BaseClassifier):
         Number of threads to use for the transform.
 
 
-    Reference
-    ---------
+    References
+    ----------
     .. [1] Mbouopda, Michael Franklin, and Engelbert Mephu Nguifo.
-    "Scalable and accurate subsequence transform for time series classification."
-    Pattern Recognition 147 (2023): 110121.
-    https://www.sciencedirect.com/science/article/abs/pii/S003132032300818X,
-    https://uca.hal.science/hal-03087686/document
+           "Scalable and accurate subsequence transform for time series classification."
+           Pattern Recognition 147 (2023): 110121.
+           https://www.sciencedirect.com/science/article/abs/pii/S003132032300818X,
+           https://uca.hal.science/hal-03087686/document
 
     Examples
     --------
@@ -92,8 +92,8 @@ def _fit(self, X, y):
         y: array-like or list
             The class values for X.
 
-        Return
-        ------
+        Returns
+        -------
         self : SASTClassifier
             This pipeline classifier
 
@@ -129,8 +129,8 @@ def _predict(self, X):
         X: np.ndarray shape (n_cases, n_channels, n_timepoints)
             The training input samples.
 
-        Return
-        ------
+        Returns
+        -------
         array-like or list
             Predicted class labels.
         """
@@ -144,8 +144,8 @@ def _predict_proba(self, X):
         X: np.ndarray shape (n_cases, n_channels, n_timepoints)
             The training input samples.
 
-        Return
-        ------
+        Returns
+        -------
         dists : np.ndarray shape (n_cases, n_timepoints)
             Predicted class probabilities.
         """

aeon/clustering/_kernel_k_means.py

@@ -64,11 +64,11 @@ class TimeSeriesKernelKMeans(BaseClusterer):
     n_iter_: int
         Number of iterations run.
 
-    Reference
-    ---------
-        .. [1] Kernel k-means, Spectral Clustering and Normalized Cuts. Inderjit S.
-        Dhillon, Yuqiang Guan, Brian Kulis. KDD 2004.
-        .. [2] Fast Global Alignment Kernels. Marco Cuturi. ICML 2011.
+    References
+    ----------
+    .. [1] Kernel k-means, Spectral Clustering and Normalized Cuts. Inderjit S.
+           Dhillon, Yuqiang Guan, Brian Kulis. KDD 2004.
+    .. [2] Fast Global Alignment Kernels. Marco Cuturi. ICML 2011.
 
     Examples
     --------

aeon/datasets/_data_loaders.py

@@ -977,8 +977,8 @@ def load_forecasting(name, extract_path=None, return_metadata=False):
         If a dataset name that does not exist on the repo is given or if a
         webpage is requested that does not exist.
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.datasets import load_forecasting
     >>> X=load_forecasting("m1_yearly_dataset") # doctest: +SKIP
     """
@@ -1134,8 +1134,8 @@ def load_regression(
         'problemname',timestamps, missing,univariate,equallength.
         targetlabel should be true, and classlabel false
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.datasets import load_regression
     >>> X, y=load_regression("FloodModeling1") # doctest: +SKIP
     """

aeon/datasets/_single_problem_loaders.py

@@ -950,8 +950,8 @@ def load_solar(
     api_version : string or None, default="v4"
         API version to call. If None then a stored sample of the data is loaded.
 
-    Return
-    ------
+    Returns
+    -------
     pd.Series
 
     References

aeon/forecasting/vecm.py

@@ -57,8 +57,8 @@ class VECM(_StatsModelsAdapter):
         2D ndarray/pd.DataFrame of size (any, neqs)
         Forcasted value of exog_coint
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.forecasting.vecm import VECM
     >>> from aeon.forecasting.model_selection import temporal_train_test_split
     >>> from aeon.forecasting.base import ForecastingHorizon

aeon/segmentation/_igts.py

@@ -154,8 +154,8 @@ class _IGTS:
        Pervasive and Mobile Computing, 38, 92-109, (2017).
        https://www.sciencedirect.com/science/article/abs/pii/S1574119217300081
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.testing.utils.data_gen import piecewise_normal_multivariate
     >>> from sklearn.preprocessing import MinMaxScaler
     >>> from aeon.segmentation import InformationGainSegmenter

aeon/transformations/collection/interpolate.py

@@ -19,8 +19,8 @@ class TSInterpolator(BaseCollectionTransformer):
     ----------
     length : integer, the length of time series to resize to.
 
-    Example
-    -------
+    Examples
+    --------
     >>> import numpy as np
     >>> from aeon.transformations.collection.interpolate import TSInterpolator
     >>> # Unequal length collection of time series

aeon/transformations/collection/pad.py

@@ -36,8 +36,8 @@ class PaddingTransformer(BaseCollectionTransformer):
         each time series. Supported statistic strings are "mean", "median", "max",
         "min".
 
-    Example
-    -------
+    Examples
+    --------
     >>> from aeon.transformations.collection import PaddingTransformer
     >>> import numpy as np
     >>> X = []

aeon/transformations/collection/shapelet_based/_rsast.py

@@ -66,8 +66,8 @@ class RSAST(BaseCollectionTransformer):
     n_jobs : int, default -1
         Number of threads to use for the transform.
 
-    Reference
-    ---------
+    References
+    ----------
     .. [1] Varela, N. R., Mbouopda, M. F., & Nguifo, E. M. (2023).
     RSAST: Sampling Shapelets for Time Series Classification.
     https://hal.science/hal-04311309/
@@ -127,8 +127,8 @@ def _fit(self, X, y):
         y: array-like or list
             The class values for X.
 
-        Return
-        ------
+        Returns
+        -------
         self : RSAST
             This transformer
 
@@ -313,8 +313,8 @@ def _transform(self, X, y=None):
         y: array-like or list
             Ignored argument, interface compatibility
 
-        Return
-        ------
+        Returns
+        -------
         X_transformed: np.ndarray shape (n_cases, n_timepoints),
             The transformed data
         """