CI: Update docs for refs/heads/main (e0dcb72)

docs/features/feature_collection.html

@@ -600,11 +600,15 @@ <h1 class="title">Module <code>tsflex.features.feature_collection</code></h1>
         &#34;&#34;&#34;
         global group_indices, group_id_name, get_group_func
         group_indices = grouped_data.indices  # dict - group_id as key; indices as value
-        group_id_name = grouped_data.grouper.names  # name of the group col(s)
+        # since in future versions of pandas grouper will be deprecated
+        group_attr = &#34;_grouper&#34; if hasattr(grouped_data, &#34;_grouper&#34;) else &#34;grouper&#34;
+        group_id_name = getattr(grouped_data, group_attr).names  # name of group col(s)
         get_group_func = self._group_feat_generator(grouped_data)
 
+        # sort_output_index can be set to False, since we want to keep the same order as
+        # the group_indices
         return self._calculate_feature_list(
-            self._executor_grouped, n_jobs, show_progress, return_df, f_handler
+            self._executor_grouped, n_jobs, show_progress, return_df, False, f_handler
         )
 
     @staticmethod
@@ -756,7 +760,6 @@ <h1 class="title">Module <code>tsflex.features.feature_collection</code></h1>
             calc_result[&#34;__end&#34;] = consecutive_grouped_by_df[&#34;end&#34;]
 
             if return_df:
-                # concatenate rows
                 return calc_result
             else:
                 return [calc_result[col] for col in calc_result.columns]
@@ -795,24 +798,28 @@ <h1 class="title">Module <code>tsflex.features.feature_collection</code></h1>
     def _calculate_feature_list(
         self,
         executor: Callable[[int], pd.DataFrame],
-        n_jobs: Optional[int],
-        show_progress: Optional[bool],
-        return_df: Optional[bool],
-        f_handler: Optional[logging.FileHandler],
+        n_jobs: Union[int, None],
+        show_progress: bool,
+        return_df: bool,
+        sort_output_index: bool,
+        f_handler: logging.FileHandler,
     ) -&gt; Union[List[pd.DataFrame], pd.DataFrame]:
         &#34;&#34;&#34;Calculate the features for the given executor.
 
         Parameters
         ----------
         executor : Callable[[int], pd.DataFrame]
             The executor function that will be used to calculate the features.
-        n_jobs : Optional[int], optional
+        n_jobs : Union[int, None]
             The number of jobs to run in parallel.
-        show_progress : Optional[bool], optional
+        show_progress : bool
             Whether to show a progress bar.
-        return_df : Optional[bool], optional
+        return_df : bool
             Whether to return a DataFrame or a list of DataFrames.
-        f_handler : Optional[logging.FileHandler], optional
+        sort_output_index : bool
+            Whether to sort the output index. Note that this is only relevant when
+            `return_df` is set to `True`.
+        f_handler : logging.FileHandler
             The file handler that is used to log the function execution times.
 
         Returns
@@ -865,7 +872,13 @@ <h1 class="title">Module <code>tsflex.features.feature_collection</code></h1>
 
         if return_df:
             # Concatenate &amp; sort the columns
-            df = pd.concat(calculated_feature_list, axis=1, join=&#34;outer&#34;, copy=False)
+            df = pd.concat(
+                calculated_feature_list,
+                axis=1,
+                join=&#34;outer&#34;,
+                copy=False,
+                sort=sort_output_index,
+            )
             return df.reindex(sorted(df.columns), axis=1)
         else:
             return calculated_feature_list
@@ -1201,6 +1214,9 @@ <h1 class="title">Module <code>tsflex.features.feature_collection</code></h1>
                     f_handler=f_handler,
                 )
 
+        # Sort output index if segment indices are not provided
+        sort_output_index = segment_start_idxs is None and segment_end_idxs is None
+
         # Convert to numpy array (if necessary)
         if segment_start_idxs is not None:
             segment_start_idxs = FeatureCollection._process_segment_idxs(
@@ -1282,7 +1298,12 @@ <h1 class="title">Module <code>tsflex.features.feature_collection</code></h1>
         )
 
         return self._calculate_feature_list(
-            self._executor_stroll, n_jobs, show_progress, return_df, f_handler
+            self._executor_stroll,
+            n_jobs,
+            show_progress,
+            return_df,
+            sort_output_index,
+            f_handler,
         )
 
     def serialize(self, file_path: Union[str, Path]):
@@ -1937,11 +1958,15 @@ <h2 class="section-title" id="header-classes">Classes</h2>
         &#34;&#34;&#34;
         global group_indices, group_id_name, get_group_func
         group_indices = grouped_data.indices  # dict - group_id as key; indices as value
-        group_id_name = grouped_data.grouper.names  # name of the group col(s)
+        # since in future versions of pandas grouper will be deprecated
+        group_attr = &#34;_grouper&#34; if hasattr(grouped_data, &#34;_grouper&#34;) else &#34;grouper&#34;
+        group_id_name = getattr(grouped_data, group_attr).names  # name of group col(s)
         get_group_func = self._group_feat_generator(grouped_data)
 
+        # sort_output_index can be set to False, since we want to keep the same order as
+        # the group_indices
         return self._calculate_feature_list(
-            self._executor_grouped, n_jobs, show_progress, return_df, f_handler
+            self._executor_grouped, n_jobs, show_progress, return_df, False, f_handler
         )
 
     @staticmethod
@@ -2093,7 +2118,6 @@ <h2 class="section-title" id="header-classes">Classes</h2>
             calc_result[&#34;__end&#34;] = consecutive_grouped_by_df[&#34;end&#34;]
 
             if return_df:
-                # concatenate rows
                 return calc_result
             else:
                 return [calc_result[col] for col in calc_result.columns]
@@ -2132,24 +2156,28 @@ <h2 class="section-title" id="header-classes">Classes</h2>
     def _calculate_feature_list(
         self,
         executor: Callable[[int], pd.DataFrame],
-        n_jobs: Optional[int],
-        show_progress: Optional[bool],
-        return_df: Optional[bool],
-        f_handler: Optional[logging.FileHandler],
+        n_jobs: Union[int, None],
+        show_progress: bool,
+        return_df: bool,
+        sort_output_index: bool,
+        f_handler: logging.FileHandler,
     ) -&gt; Union[List[pd.DataFrame], pd.DataFrame]:
         &#34;&#34;&#34;Calculate the features for the given executor.
 
         Parameters
         ----------
         executor : Callable[[int], pd.DataFrame]
             The executor function that will be used to calculate the features.
-        n_jobs : Optional[int], optional
+        n_jobs : Union[int, None]
             The number of jobs to run in parallel.
-        show_progress : Optional[bool], optional
+        show_progress : bool
             Whether to show a progress bar.
-        return_df : Optional[bool], optional
+        return_df : bool
             Whether to return a DataFrame or a list of DataFrames.
-        f_handler : Optional[logging.FileHandler], optional
+        sort_output_index : bool
+            Whether to sort the output index. Note that this is only relevant when
+            `return_df` is set to `True`.
+        f_handler : logging.FileHandler
             The file handler that is used to log the function execution times.
 
         Returns
@@ -2202,7 +2230,13 @@ <h2 class="section-title" id="header-classes">Classes</h2>
 
         if return_df:
             # Concatenate &amp; sort the columns
-            df = pd.concat(calculated_feature_list, axis=1, join=&#34;outer&#34;, copy=False)
+            df = pd.concat(
+                calculated_feature_list,
+                axis=1,
+                join=&#34;outer&#34;,
+                copy=False,
+                sort=sort_output_index,
+            )
             return df.reindex(sorted(df.columns), axis=1)
         else:
             return calculated_feature_list
@@ -2538,6 +2572,9 @@ <h2 class="section-title" id="header-classes">Classes</h2>
                     f_handler=f_handler,
                 )
 
+        # Sort output index if segment indices are not provided
+        sort_output_index = segment_start_idxs is None and segment_end_idxs is None
+
         # Convert to numpy array (if necessary)
         if segment_start_idxs is not None:
             segment_start_idxs = FeatureCollection._process_segment_idxs(
@@ -2619,7 +2656,12 @@ <h2 class="section-title" id="header-classes">Classes</h2>
         )
 
         return self._calculate_feature_list(
-            self._executor_stroll, n_jobs, show_progress, return_df, f_handler
+            self._executor_stroll,
+            n_jobs,
+            show_progress,
+            return_df,
+            sort_output_index,
+            f_handler,
         )
 
     def serialize(self, file_path: Union[str, Path]):
@@ -3234,6 +3276,9 @@ <h2 id="raises">Raises</h2>
                 f_handler=f_handler,
             )
 
+    # Sort output index if segment indices are not provided
+    sort_output_index = segment_start_idxs is None and segment_end_idxs is None
+
     # Convert to numpy array (if necessary)
     if segment_start_idxs is not None:
         segment_start_idxs = FeatureCollection._process_segment_idxs(
@@ -3315,7 +3360,12 @@ <h2 id="raises">Raises</h2>
     )
 
     return self._calculate_feature_list(
-        self._executor_stroll, n_jobs, show_progress, return_df, f_handler
+        self._executor_stroll,
+        n_jobs,
+        show_progress,
+        return_df,
+        sort_output_index,
+        f_handler,
     )</code></pre>
 </details>
 <div class="desc"><p>Calculate features on the passed data.</p>

docs/features/index.html

@@ -1638,11 +1638,15 @@ <h2 id="note">Note</h2>
         &#34;&#34;&#34;
         global group_indices, group_id_name, get_group_func
         group_indices = grouped_data.indices  # dict - group_id as key; indices as value
-        group_id_name = grouped_data.grouper.names  # name of the group col(s)
+        # since in future versions of pandas grouper will be deprecated
+        group_attr = &#34;_grouper&#34; if hasattr(grouped_data, &#34;_grouper&#34;) else &#34;grouper&#34;
+        group_id_name = getattr(grouped_data, group_attr).names  # name of group col(s)
         get_group_func = self._group_feat_generator(grouped_data)
 
+        # sort_output_index can be set to False, since we want to keep the same order as
+        # the group_indices
         return self._calculate_feature_list(
-            self._executor_grouped, n_jobs, show_progress, return_df, f_handler
+            self._executor_grouped, n_jobs, show_progress, return_df, False, f_handler
         )
 
     @staticmethod
@@ -1794,7 +1798,6 @@ <h2 id="note">Note</h2>
             calc_result[&#34;__end&#34;] = consecutive_grouped_by_df[&#34;end&#34;]
 
             if return_df:
-                # concatenate rows
                 return calc_result
             else:
                 return [calc_result[col] for col in calc_result.columns]
@@ -1833,24 +1836,28 @@ <h2 id="note">Note</h2>
     def _calculate_feature_list(
         self,
         executor: Callable[[int], pd.DataFrame],
-        n_jobs: Optional[int],
-        show_progress: Optional[bool],
-        return_df: Optional[bool],
-        f_handler: Optional[logging.FileHandler],
+        n_jobs: Union[int, None],
+        show_progress: bool,
+        return_df: bool,
+        sort_output_index: bool,
+        f_handler: logging.FileHandler,
     ) -&gt; Union[List[pd.DataFrame], pd.DataFrame]:
         &#34;&#34;&#34;Calculate the features for the given executor.
 
         Parameters
         ----------
         executor : Callable[[int], pd.DataFrame]
             The executor function that will be used to calculate the features.
-        n_jobs : Optional[int], optional
+        n_jobs : Union[int, None]
             The number of jobs to run in parallel.
-        show_progress : Optional[bool], optional
+        show_progress : bool
             Whether to show a progress bar.
-        return_df : Optional[bool], optional
+        return_df : bool
             Whether to return a DataFrame or a list of DataFrames.
-        f_handler : Optional[logging.FileHandler], optional
+        sort_output_index : bool
+            Whether to sort the output index. Note that this is only relevant when
+            `return_df` is set to `True`.
+        f_handler : logging.FileHandler
             The file handler that is used to log the function execution times.
 
         Returns
@@ -1903,7 +1910,13 @@ <h2 id="note">Note</h2>
 
         if return_df:
             # Concatenate &amp; sort the columns
-            df = pd.concat(calculated_feature_list, axis=1, join=&#34;outer&#34;, copy=False)
+            df = pd.concat(
+                calculated_feature_list,
+                axis=1,
+                join=&#34;outer&#34;,
+                copy=False,
+                sort=sort_output_index,
+            )
             return df.reindex(sorted(df.columns), axis=1)
         else:
             return calculated_feature_list
@@ -2239,6 +2252,9 @@ <h2 id="note">Note</h2>
                     f_handler=f_handler,
                 )
 
+        # Sort output index if segment indices are not provided
+        sort_output_index = segment_start_idxs is None and segment_end_idxs is None
+
         # Convert to numpy array (if necessary)
         if segment_start_idxs is not None:
             segment_start_idxs = FeatureCollection._process_segment_idxs(
@@ -2320,7 +2336,12 @@ <h2 id="note">Note</h2>
         )
 
         return self._calculate_feature_list(
-            self._executor_stroll, n_jobs, show_progress, return_df, f_handler
+            self._executor_stroll,
+            n_jobs,
+            show_progress,
+            return_df,
+            sort_output_index,
+            f_handler,
         )
 
     def serialize(self, file_path: Union[str, Path]):
@@ -2935,6 +2956,9 @@ <h2 id="raises">Raises</h2>
                 f_handler=f_handler,
             )
 
+    # Sort output index if segment indices are not provided
+    sort_output_index = segment_start_idxs is None and segment_end_idxs is None
+
     # Convert to numpy array (if necessary)
     if segment_start_idxs is not None:
         segment_start_idxs = FeatureCollection._process_segment_idxs(
@@ -3016,7 +3040,12 @@ <h2 id="raises">Raises</h2>
     )
 
     return self._calculate_feature_list(
-        self._executor_stroll, n_jobs, show_progress, return_df, f_handler
+        self._executor_stroll,
+        n_jobs,
+        show_progress,
+        return_df,
+        sort_output_index,
+        f_handler,
     )</code></pre>
 </details>
 <div class="desc"><p>Calculate features on the passed data.</p>

docs/features/segmenter/strided_rolling.html

@@ -626,6 +626,10 @@ <h1 class="title">Module <code>tsflex.features.segmenter.strided_rolling</code><
     ) -&gt; pd.Index:
         assert start_idxs.dtype.type == np.datetime64
         assert end_idxs.dtype.type == np.datetime64
+        if not len(start_idxs):  # to fix &#34;datetime64 values must have a unit specified&#34;
+            assert not len(end_idxs)
+            start_idxs = start_idxs.astype(&#34;datetime64[ns]&#34;)
+            end_idxs = end_idxs.astype(&#34;datetime64[ns]&#34;)
         start_idxs = pd.to_datetime(start_idxs, utc=True).tz_convert(self._tz_index)
         end_idxs = pd.to_datetime(end_idxs, utc=True).tz_convert(self._tz_index)
         return super()._get_output_index(start_idxs, end_idxs, name)
@@ -1803,6 +1807,10 @@ <h3>Methods</h3>
     ) -&gt; pd.Index:
         assert start_idxs.dtype.type == np.datetime64
         assert end_idxs.dtype.type == np.datetime64
+        if not len(start_idxs):  # to fix &#34;datetime64 values must have a unit specified&#34;
+            assert not len(end_idxs)
+            start_idxs = start_idxs.astype(&#34;datetime64[ns]&#34;)
+            end_idxs = end_idxs.astype(&#34;datetime64[ns]&#34;)
         start_idxs = pd.to_datetime(start_idxs, utc=True).tz_convert(self._tz_index)
         end_idxs = pd.to_datetime(end_idxs, utc=True).tz_convert(self._tz_index)
         return super()._get_output_index(start_idxs, end_idxs, name)