eegdash 0.3.3.dev61__py3-none-any.whl → 0.5.0.dev180784713__py3-none-any.whl

eegdash/features/datasets.py

@@ -3,7 +3,6 @@ from __future__ import annotations
 import json
 import os
 import shutil
-import warnings
 from collections.abc import Callable
 from typing import Dict, List
 
@@ -17,22 +16,43 @@ from braindecode.datasets.base import (
     _create_description,
 )
 
+from ..logging import logger
+
+__all__ = [
+    "FeaturesDataset",
+    "FeaturesConcatDataset",
+]
+
 
 class FeaturesDataset(EEGWindowsDataset):
-    """Returns samples from a pandas DataFrame object along with a target.
+    """A dataset of features extracted from EEG windows.
 
-    Dataset which serves samples from a pandas DataFrame object along with a
-    target. The target is unique for the dataset, and is obtained through the
-    `description` attribute.
+    This class holds features in a pandas DataFrame and provides an interface
+    compatible with braindecode's dataset structure. Each row in the feature
+    DataFrame corresponds to a single sample (e.g., an EEG window).
 
     Parameters
     ----------
-    features : a pandas DataFrame
-        Tabular data.
-    description : dict | pandas.Series | None
-        Holds additional description about the continuous signal / subject.
-    transform : callable | None
-        On-the-fly transform applied to the example before it is returned.
+    features : pandas.DataFrame
+        A DataFrame where each row is a sample and each column is a feature.
+    metadata : pandas.DataFrame, optional
+        A DataFrame containing metadata for each sample, indexed consistently
+        with `features`. Must include columns 'i_window_in_trial',
+        'i_start_in_trial', 'i_stop_in_trial', and 'target'.
+    description : dict or pandas.Series, optional
+        Additional high-level information about the dataset (e.g., subject ID).
+    transform : callable, optional
+        A function or transform to apply to the feature data on-the-fly.
+    raw_info : dict, optional
+        Information about the original raw recording, for provenance.
+    raw_preproc_kwargs : dict, optional
+        Keyword arguments used for preprocessing the raw data.
+    window_kwargs : dict, optional
+        Keyword arguments used for windowing the data.
+    window_preproc_kwargs : dict, optional
+        Keyword arguments used for preprocessing the windowed data.
+    features_kwargs : dict, optional
+        Keyword arguments used for feature extraction.
 
     """
 
@@ -64,7 +84,21 @@ class FeaturesDataset(EEGWindowsDataset):
         ].to_numpy()
         self.y = metadata.loc[:, "target"].to_list()
 
-    def __getitem__(self, index):
+    def __getitem__(self, index: int) -> tuple[np.ndarray, int, list]:
+        """Get a single sample from the dataset.
+
+        Parameters
+        ----------
+        index : int
+            The index of the sample to retrieve.
+
+        Returns
+        -------
+        tuple
+            A tuple containing the feature vector (X), the target (y), and the
+            cropping indices.
+
+        """
         crop_inds = self.crop_inds[index].tolist()
         X = self.features.iloc[index].to_numpy()
         X = X.copy()
@@ -74,18 +108,27 @@ class FeaturesDataset(EEGWindowsDataset):
         y = self.y[index]
         return X, y, crop_inds
 
-    def __len__(self):
+    def __len__(self) -> int:
+        """Return the number of samples in the dataset.
+
+        Returns
+        -------
+        int
+            The total number of feature samples.
+
+        """
         return len(self.features.index)
 
 
 def _compute_stats(
     ds: FeaturesDataset,
-    return_count=False,
-    return_mean=False,
-    return_var=False,
-    ddof=1,
-    numeric_only=False,
-):
+    return_count: bool = False,
+    return_mean: bool = False,
+    return_var: bool = False,
+    ddof: int = 1,
+    numeric_only: bool = False,
+) -> tuple:
+    """Compute statistics for a single :class:`~eegdash.features.datasets.FeaturesDataset`."""
     res = []
     if return_count:
         res.append(ds.features.count(numeric_only=numeric_only))
@@ -96,7 +139,14 @@ def _compute_stats(
     return tuple(res)
 
 
-def _pooled_var(counts, means, variances, ddof, ddof_in=None):
+def _pooled_var(
+    counts: np.ndarray,
+    means: np.ndarray,
+    variances: np.ndarray,
+    ddof: int,
+    ddof_in: int | None = None,
+) -> tuple[np.ndarray, np.ndarray, np.ndarray]:
+    """Compute pooled variance across multiple datasets."""
     if ddof_in is None:
         ddof_in = ddof
     count = counts.sum(axis=0)
@@ -109,17 +159,20 @@ def _pooled_var(counts, means, variances, ddof, ddof_in=None):
 
 
 class FeaturesConcatDataset(BaseConcatDataset):
-    """A base class for concatenated datasets.
+    """A concatenated dataset of :class:`~eegdash.features.datasets.FeaturesDataset` objects.
 
-    Holds either mne.Raw or mne.Epoch in self.datasets and has
-    a pandas DataFrame with additional description.
+    This class holds a list of :class:`~eegdash.features.datasets.FeaturesDataset` instances and allows
+    them to be treated as a single, larger dataset. It provides methods for
+
+    splitting, saving, and performing DataFrame-like operations (e.g., `mean`,
+    `var`, `fillna`) across all contained datasets.
 
     Parameters
     ----------
-    list_of_ds : list
-        list of BaseDataset, BaseConcatDataset or WindowsDataset
-    target_transform : callable | None
-        Optional function to call on targets before returning them.
+    list_of_ds : list of ~eegdash.features.datasets.FeaturesDataset
+        A list of :class:`~eegdash.features.datasets.FeaturesDataset` objects to concatenate.
+    target_transform : callable, optional
+        A function to apply to the target values before they are returned.
 
     """
 
@@ -139,26 +192,28 @@ class FeaturesConcatDataset(BaseConcatDataset):
         self,
         by: str | list[int] | list[list[int]] | dict[str, list[int]],
     ) -> dict[str, FeaturesConcatDataset]:
-        """Split the dataset based on information listed in its description.
+        """Split the dataset into subsets.
 
-        The format could be based on a DataFrame or based on indices.
+        The splitting can be done based on a column in the description
+        DataFrame or by providing explicit indices for each split.
 
         Parameters
         ----------
-        by : str | list | dict
-            If ``by`` is a string, splitting is performed based on the
-            description DataFrame column with this name.
-            If ``by`` is a (list of) list of integers, the position in the first
-            list corresponds to the split id and the integers to the
-            datapoints of that split.
-            If a dict then each key will be used in the returned
-            splits dict and each value should be a list of int.
+        by : str or list or dict
+            - If a string, splits are created for each unique value in the
+              description column `by`.
+            - If a list of integers, a single split is created containing the
+              datasets at the specified indices.
+            - If a list of lists of integers, multiple splits are created, one
+              for each sublist of indices.
+            - If a dictionary, keys are used as split names and values are
+              lists of dataset indices.
 
         Returns
         -------
-        splits : dict
-            A dictionary with the name of the split (a string) as key and the
-            dataset as value.
+        dict[str, ~eegdash.features.datasets.FeaturesConcatDataset]
+            A dictionary where keys are split names and values are the new
+            :class:`~eegdash.features.datasets.FeaturesConcatDataset` subsets.
 
         """
         if isinstance(by, str):
@@ -183,14 +238,22 @@ class FeaturesConcatDataset(BaseConcatDataset):
         }
 
     def get_metadata(self) -> pd.DataFrame:
-        """Concatenate the metadata and description of the wrapped Epochs.
+        """Get the metadata of all datasets as a single DataFrame.
+
+        Concatenates the metadata from all contained datasets and adds columns
+        from their `description` attributes.
 
         Returns
         -------
-        metadata : pd.DataFrame
-            DataFrame containing as many rows as there are windows in the
-            BaseConcatDataset, with the metadata and description information
-            for each window.
+        pandas.DataFrame
+            A DataFrame containing the metadata for every sample in the
+            concatenated dataset.
+
+        Raises
+        ------
+        TypeError
+            If any of the contained datasets is not a
+            :class:`~eegdash.features.datasets.FeaturesDataset`.
 
         """
         if not all([isinstance(ds, FeaturesDataset) for ds in self.datasets]):
@@ -201,60 +264,59 @@ class FeaturesConcatDataset(BaseConcatDataset):
 
         all_dfs = list()
         for ds in self.datasets:
-            df = ds.metadata
+            df = ds.metadata.copy()
             for k, v in ds.description.items():
                 df[k] = v
             all_dfs.append(df)
 
         return pd.concat(all_dfs)
 
-    def save(self, path: str, overwrite: bool = False, offset: int = 0):
-        """Save datasets to files by creating one subdirectory for each dataset:
-        path/
-            0/
-                0-feat.parquet
-                metadata_df.pkl
-                description.json
-                raw-info.fif (if raw info was saved)
-                raw_preproc_kwargs.json (if raws were preprocessed)
-                window_kwargs.json (if this is a windowed dataset)
-                window_preproc_kwargs.json  (if windows were preprocessed)
-                features_kwargs.json
-            1/
-                1-feat.parquet
-                metadata_df.pkl
-                description.json
-                raw-info.fif (if raw info was saved)
-                raw_preproc_kwargs.json (if raws were preprocessed)
-                window_kwargs.json (if this is a windowed dataset)
-                window_preproc_kwargs.json  (if windows were preprocessed)
-                features_kwargs.json
+    def save(self, path: str, overwrite: bool = False, offset: int = 0) -> None:
+        """Save the concatenated dataset to a directory.
+
+        Creates a directory structure where each contained dataset is saved in
+        its own numbered subdirectory.
+
+        .. code-block::
+
+            path/
+                0/
+                    0-feat.parquet
+                    metadata_df.pkl
+                    description.json
+                    ...
+                1/
+                    1-feat.parquet
+                    ...
 
         Parameters
         ----------
         path : str
-            Directory in which subdirectories are created to store
-             -feat.parquet and .json files to.
-        overwrite : bool
-            Whether to delete old subdirectories that will be saved to in this
-            call.
-        offset : int
-            If provided, the integer is added to the id of the dataset in the
-            concat. This is useful in the setting of very large datasets, where
-            one dataset has to be processed and saved at a time to account for
-            its original position.
+            The directory where the dataset will be saved.
+        overwrite : bool, default False
+            If True, any existing subdirectories that conflict with the new
+            ones will be removed.
+        offset : int, default 0
+            An integer to add to the subdirectory names. Useful for saving
+            datasets in chunks.
+
+        Raises
+        ------
+        ValueError
+            If the dataset is empty.
+        FileExistsError
+            If a subdirectory already exists and `overwrite` is False.
 
         """
         if len(self.datasets) == 0:
             raise ValueError("Expect at least one dataset")
         path_contents = os.listdir(path)
-        n_sub_dirs = len([os.path.isdir(e) for e in path_contents])
+        n_sub_dirs = len([os.path.isdir(os.path.join(path, e)) for e in path_contents])
         for i_ds, ds in enumerate(self.datasets):
-            # remove subdirectory from list of untouched files / subdirectories
-            if str(i_ds + offset) in path_contents:
-                path_contents.remove(str(i_ds + offset))
-            # save_dir/i_ds/
-            sub_dir = os.path.join(path, str(i_ds + offset))
+            sub_dir_name = str(i_ds + offset)
+            if sub_dir_name in path_contents:
+                path_contents.remove(sub_dir_name)
+            sub_dir = os.path.join(path, sub_dir_name)
             if os.path.exists(sub_dir):
                 if overwrite:
                     shutil.rmtree(sub_dir)
@@ -264,56 +326,59 @@ class FeaturesConcatDataset(BaseConcatDataset):
                         f" a different directory, set overwrite=True, or "
                         f"resolve manually."
                     )
-            # save_dir/{i_ds+offset}/
             os.makedirs(sub_dir)
-            # save_dir/{i_ds+offset}/{i_ds+offset}-feat.parquet
             self._save_features(sub_dir, ds, i_ds, offset)
-            # save_dir/{i_ds+offset}/metadata_df.pkl
             self._save_metadata(sub_dir, ds)
-            # save_dir/{i_ds+offset}/description.json
             self._save_description(sub_dir, ds.description)
-            # save_dir/{i_ds+offset}/raw-info.fif
             self._save_raw_info(sub_dir, ds)
-            # save_dir/{i_ds+offset}/raw_preproc_kwargs.json
-            # save_dir/{i_ds+offset}/window_kwargs.json
-            # save_dir/{i_ds+offset}/window_preproc_kwargs.json
-            # save_dir/{i_ds+offset}/features_kwargs.json
             self._save_kwargs(sub_dir, ds)
-        if overwrite:
-            # the following will be True for all datasets preprocessed and
-            # stored in parallel with braindecode.preprocessing.preprocess
-            if i_ds + 1 + offset < n_sub_dirs:
-                warnings.warn(
-                    f"The number of saved datasets ({i_ds + 1 + offset}) "
-                    f"does not match the number of existing "
-                    f"subdirectories ({n_sub_dirs}). You may now "
-                    f"encounter a mix of differently preprocessed "
-                    f"datasets!",
-                    UserWarning,
-                )
-        # if path contains files or directories that were not touched, raise
-        # warning
+        if overwrite and i_ds + 1 + offset < n_sub_dirs:
+            logger.warning(
+                f"The number of saved datasets ({i_ds + 1 + offset}) "
+                f"does not match the number of existing "
+                f"subdirectories ({n_sub_dirs}). You may now "
+                f"encounter a mix of differently preprocessed "
+                f"datasets!",
+                UserWarning,
+            )
         if path_contents:
-            warnings.warn(
+            logger.warning(
                 f"Chosen directory {path} contains other "
                 f"subdirectories or files {path_contents}."
             )
 
     @staticmethod
-    def _save_features(sub_dir, ds, i_ds, offset):
+    def _save_features(sub_dir: str, ds: FeaturesDataset, i_ds: int, offset: int):
+        """Save the feature DataFrame to a Parquet file."""
         parquet_file_name = f"{i_ds + offset}-feat.parquet"
         parquet_file_path = os.path.join(sub_dir, parquet_file_name)
         ds.features.to_parquet(parquet_file_path)
 
     @staticmethod
-    def _save_raw_info(sub_dir, ds):
-        if hasattr(ds, "raw_info"):
+    def _save_metadata(sub_dir: str, ds: FeaturesDataset):
+        """Save the metadata DataFrame to a pickle file."""
+        metadata_file_name = "metadata_df.pkl"
+        metadata_file_path = os.path.join(sub_dir, metadata_file_name)
+        ds.metadata.to_pickle(metadata_file_path)
+
+    @staticmethod
+    def _save_description(sub_dir: str, description: pd.Series):
+        """Save the description Series to a JSON file."""
+        desc_file_name = "description.json"
+        desc_file_path = os.path.join(sub_dir, desc_file_name)
+        description.to_json(desc_file_path)
+
+    @staticmethod
+    def _save_raw_info(sub_dir: str, ds: FeaturesDataset):
+        """Save the raw info dictionary to a FIF file if it exists."""
+        if hasattr(ds, "raw_info") and ds.raw_info is not None:
             fif_file_name = "raw-info.fif"
             fif_file_path = os.path.join(sub_dir, fif_file_name)
-            ds.raw_info.save(fif_file_path)
+            ds.raw_info.save(fif_file_path, overwrite=True)
 
     @staticmethod
-    def _save_kwargs(sub_dir, ds):
+    def _save_kwargs(sub_dir: str, ds: FeaturesDataset):
+        """Save various keyword argument dictionaries to JSON files."""
         for kwargs_name in [
             "raw_preproc_kwargs",
             "window_kwargs",
@@ -321,10 +386,10 @@ class FeaturesConcatDataset(BaseConcatDataset):
             "features_kwargs",
         ]:
             if hasattr(ds, kwargs_name):
-                kwargs_file_name = ".".join([kwargs_name, "json"])
-                kwargs_file_path = os.path.join(sub_dir, kwargs_file_name)
                 kwargs = getattr(ds, kwargs_name)
                 if kwargs is not None:
+                    kwargs_file_name = ".".join([kwargs_name, "json"])
+                    kwargs_file_path = os.path.join(sub_dir, kwargs_file_name)
                     with open(kwargs_file_path, "w") as f:
                         json.dump(kwargs, f)
 
@@ -333,7 +398,25 @@ class FeaturesConcatDataset(BaseConcatDataset):
         include_metadata: bool | str | List[str] = False,
         include_target: bool = False,
         include_crop_inds: bool = False,
-    ):
+    ) -> pd.DataFrame:
+        """Convert the dataset to a single pandas DataFrame.
+
+        Parameters
+        ----------
+        include_metadata : bool or str or list of str, default False
+            If True, include all metadata columns. If a string or list of
+            strings, include only the specified metadata columns.
+        include_target : bool, default False
+            If True, include the 'target' column.
+        include_crop_inds : bool, default False
+            If True, include window cropping index columns.
+
+        Returns
+        -------
+        pandas.DataFrame
+            A DataFrame containing the features and requested metadata.
+
+        """
         if (
             not isinstance(include_metadata, bool)
             or include_metadata
@@ -342,7 +425,7 @@ class FeaturesConcatDataset(BaseConcatDataset):
             include_dataset = False
             if isinstance(include_metadata, bool) and include_metadata:
                 include_dataset = True
-                cols = self.datasets[0].metadata.columns
+                cols = self.datasets[0].metadata.columns.tolist()
             else:
                 cols = include_metadata
                 if isinstance(cols, bool) and not cols:
@@ -351,13 +434,14 @@ class FeaturesConcatDataset(BaseConcatDataset):
                     cols = [cols]
                 cols = set(cols)
                 if include_crop_inds:
-                    cols = {
-                        "i_dataset",
-                        "i_window_in_trial",
-                        "i_start_in_trial",
-                        "i_stop_in_trial",
-                        *cols,
-                    }
+                    cols.update(
+                        {
+                            "i_dataset",
+                            "i_window_in_trial",
+                            "i_start_in_trial",
+                            "i_stop_in_trial",
+                        }
+                    )
                 if include_target:
                     cols.add("target")
                 cols = list(cols)
@@ -380,10 +464,26 @@ class FeaturesConcatDataset(BaseConcatDataset):
             dataframes = [ds.features for ds in self.datasets]
         return pd.concat(dataframes, axis=0, ignore_index=True)
 
-    def _numeric_columns(self):
+    def _numeric_columns(self) -> pd.Index:
+        """Get the names of numeric columns from the feature DataFrames."""
         return self.datasets[0].features.select_dtypes(include=np.number).columns
 
-    def count(self, numeric_only=False, n_jobs=1):
+    def count(self, numeric_only: bool = False, n_jobs: int = 1) -> pd.Series:
+        """Count non-NA cells for each feature column.
+
+        Parameters
+        ----------
+        numeric_only : bool, default False
+            Include only float, int, boolean columns.
+        n_jobs : int, default 1
+            Number of jobs to run in parallel.
+
+        Returns
+        -------
+        pandas.Series
+            The count of non-NA cells for each column.
+
+        """
         stats = Parallel(n_jobs)(
             delayed(_compute_stats)(ds, return_count=True, numeric_only=numeric_only)
             for ds in self.datasets
@@ -392,7 +492,22 @@ class FeaturesConcatDataset(BaseConcatDataset):
         count = counts.sum(axis=0)
         return pd.Series(count, index=self._numeric_columns())
 
-    def mean(self, numeric_only=False, n_jobs=1):
+    def mean(self, numeric_only: bool = False, n_jobs: int = 1) -> pd.Series:
+        """Compute the mean for each feature column.
+
+        Parameters
+        ----------
+        numeric_only : bool, default False
+            Include only float, int, boolean columns.
+        n_jobs : int, default 1
+            Number of jobs to run in parallel.
+
+        Returns
+        -------
+        pandas.Series
+            The mean of each column.
+
+        """
         stats = Parallel(n_jobs)(
             delayed(_compute_stats)(
                 ds, return_count=True, return_mean=True, numeric_only=numeric_only
@@ -404,7 +519,26 @@ class FeaturesConcatDataset(BaseConcatDataset):
         mean = np.sum((counts / count) * means, axis=0)
         return pd.Series(mean, index=self._numeric_columns())
 
-    def var(self, ddof=1, numeric_only=False, n_jobs=1):
+    def var(
+        self, ddof: int = 1, numeric_only: bool = False, n_jobs: int = 1
+    ) -> pd.Series:
+        """Compute the variance for each feature column.
+
+        Parameters
+        ----------
+        ddof : int, default 1
+            Delta Degrees of Freedom. The divisor used in calculations is N - ddof.
+        numeric_only : bool, default False
+            Include only float, int, boolean columns.
+        n_jobs : int, default 1
+            Number of jobs to run in parallel.
+
+        Returns
+        -------
+        pandas.Series
+            The variance of each column.
+
+        """
         stats = Parallel(n_jobs)(
             delayed(_compute_stats)(
                 ds,
@@ -424,12 +558,50 @@ class FeaturesConcatDataset(BaseConcatDataset):
         _, _, var = _pooled_var(counts, means, variances, ddof, ddof_in=0)
         return pd.Series(var, index=self._numeric_columns())
 
-    def std(self, ddof=1, numeric_only=False, eps=0, n_jobs=1):
+    def std(
+        self, ddof: int = 1, numeric_only: bool = False, eps: float = 0, n_jobs: int = 1
+    ) -> pd.Series:
+        """Compute the standard deviation for each feature column.
+
+        Parameters
+        ----------
+        ddof : int, default 1
+            Delta Degrees of Freedom.
+        numeric_only : bool, default False
+            Include only float, int, boolean columns.
+        eps : float, default 0
+            A small epsilon value to add to the variance before taking the
+            square root to avoid numerical instability.
+        n_jobs : int, default 1
+            Number of jobs to run in parallel.
+
+        Returns
+        -------
+        pandas.Series
+            The standard deviation of each column.
+
+        """
         return np.sqrt(
             self.var(ddof=ddof, numeric_only=numeric_only, n_jobs=n_jobs) + eps
         )
 
-    def zscore(self, ddof=1, numeric_only=False, eps=0, n_jobs=1):
+    def zscore(
+        self, ddof: int = 1, numeric_only: bool = False, eps: float = 0, n_jobs: int = 1
+    ) -> None:
+        """Apply z-score normalization to numeric columns in-place.
+
+        Parameters
+        ----------
+        ddof : int, default 1
+            Delta Degrees of Freedom for variance calculation.
+        numeric_only : bool, default False
+            Include only float, int, boolean columns.
+        eps : float, default 0
+            Epsilon for numerical stability.
+        n_jobs : int, default 1
+            Number of jobs to run in parallel for statistics computation.
+
+        """
         stats = Parallel(n_jobs)(
             delayed(_compute_stats)(
                 ds,
@@ -449,10 +621,13 @@ class FeaturesConcatDataset(BaseConcatDataset):
         _, mean, var = _pooled_var(counts, means, variances, ddof, ddof_in=0)
         std = np.sqrt(var + eps)
         for ds in self.datasets:
-            ds.features = (ds.features - mean) / std
+            ds.features.loc[:, self._numeric_columns()] = (
+                ds.features.loc[:, self._numeric_columns()] - mean
+            ) / std
 
     @staticmethod
-    def _enforce_inplace_operations(func_name, kwargs):
+    def _enforce_inplace_operations(func_name: str, kwargs: dict):
+        """Raise an error if 'inplace=False' is passed to a method."""
         if "inplace" in kwargs and kwargs["inplace"] is False:
             raise ValueError(
                 f"{func_name} only works inplace, please change "
@@ -460,33 +635,49 @@ class FeaturesConcatDataset(BaseConcatDataset):
             )
         kwargs["inplace"] = True
 
-    def fillna(self, *args, **kwargs):
+    def fillna(self, *args, **kwargs) -> None:
+        """Fill NA/NaN values in-place. See :meth:`pandas.DataFrame.fillna`."""
         FeaturesConcatDataset._enforce_inplace_operations("fillna", kwargs)
         for ds in self.datasets:
             ds.features.fillna(*args, **kwargs)
 
-    def replace(self, *args, **kwargs):
+    def replace(self, *args, **kwargs) -> None:
+        """Replace values in-place. See :meth:`pandas.DataFrame.replace`."""
         FeaturesConcatDataset._enforce_inplace_operations("replace", kwargs)
         for ds in self.datasets:
             ds.features.replace(*args, **kwargs)
 
-    def interpolate(self, *args, **kwargs):
+    def interpolate(self, *args, **kwargs) -> None:
+        """Interpolate values in-place. See :meth:`pandas.DataFrame.interpolate`."""
         FeaturesConcatDataset._enforce_inplace_operations("interpolate", kwargs)
         for ds in self.datasets:
             ds.features.interpolate(*args, **kwargs)
 
-    def dropna(self, *args, **kwargs):
+    def dropna(self, *args, **kwargs) -> None:
+        """Remove missing values in-place. See :meth:`pandas.DataFrame.dropna`."""
         FeaturesConcatDataset._enforce_inplace_operations("dropna", kwargs)
         for ds in self.datasets:
             ds.features.dropna(*args, **kwargs)
 
-    def drop(self, *args, **kwargs):
+    def drop(self, *args, **kwargs) -> None:
+        """Drop specified labels from rows or columns in-place. See :meth:`pandas.DataFrame.drop`."""
         FeaturesConcatDataset._enforce_inplace_operations("drop", kwargs)
         for ds in self.datasets:
             ds.features.drop(*args, **kwargs)
 
-    def join(self, concat_dataset: FeaturesConcatDataset, **kwargs):
+    def join(self, concat_dataset: FeaturesConcatDataset, **kwargs) -> None:
+        """Join columns with other FeaturesConcatDataset in-place.
+
+        Parameters
+        ----------
+        concat_dataset : FeaturesConcatDataset
+            The dataset to join with. Must have the same number of datasets,
+            and each corresponding dataset must have the same length.
+        **kwargs
+            Keyword arguments to pass to :meth:`pandas.DataFrame.join`.
+
+        """
         assert len(self.datasets) == len(concat_dataset.datasets)
         for ds1, ds2 in zip(self.datasets, concat_dataset.datasets):
             assert len(ds1) == len(ds2)
-            ds1.features.join(ds2, **kwargs)
+            ds1.features = ds1.features.join(ds2.features, **kwargs)