[ENH] purge todo (#995)

* switch test example for pipeline

* switch test example for pipeline

* purge todo

* purge todo

aeon/datatypes/_adapter/dask_to_pd.py

@@ -204,7 +204,6 @@ def ret(valid, msg, metadata, return_metadata):
     # check whether index is equally spaced or if there are any nans
     #   compute only if needed
     if return_metadata:
-        # todo: logic for equal spacing
         metadata["is_equally_spaced"] = True
         metadata["has_nans"] = obj.isnull().values.any().compute()
 

aeon/datatypes/_panel/_check.py

@@ -428,7 +428,6 @@ def is_nested_dataframe(obj, return_metadata=False, var_name="obj"):
         metadata["has_nans"] = _nested_dataframe_has_nans(obj)
         metadata["is_equal_length"] = not _nested_dataframe_has_unequal(obj)
 
-    # todo: this is temporary override, proper is_empty logic needs to be added
     metadata["is_empty"] = False
     metadata["is_equally_spaced"] = True
     # end hacks

aeon/datatypes/_proba/_convert.py

@@ -86,7 +86,6 @@ def convert_pred_interval_to_quantiles(y_pred, inplace=False):
     var_names = idx.get_level_values(0)
 
     # treat univariate default name
-    # todo: maybe not a good idea, remove this...
     # here because it's in the current specification
     var_names = ["Quantiles" if x == "Coverage" else x for x in var_names]
 
@@ -153,7 +152,6 @@ def convert_pred_quantiles_to_interval(y_pred, inplace=False):
     var_names = idx.get_level_values(0)
 
     # treat univariate default name
-    # todo: maybe not a good idea, remove this...
     # here because it's in the current specification
     var_names = ["Coverage" if x == "Quantiles" else x for x in var_names]
 

aeon/datatypes/tests/test_check.py

@@ -111,8 +111,6 @@ def test_check_positive(scitype, mtype, fixture_index):
     """
     # retrieve fixture for checking
     fixture = get_examples(mtype=mtype, as_scitype=scitype).get(fixture_index)
-
-    # todo: possibly remove this once all checks are defined
     check_is_defined = (mtype, scitype) in check_dict.keys()
 
     # check fixtures that exist against checks that exist
@@ -147,7 +145,6 @@ def test_check_metadata_inference(scitype, mtype, fixture_index):
         mtype=mtype, as_scitype=scitype, return_metadata=True
     ).get(fixture_index)
 
-    # todo: possibly remove this once all checks are defined
     check_is_defined = (mtype, scitype) in check_dict.keys()
     # if the examples have no metadata to them, don't test
     metadata_provided = expected_metadata is not None
@@ -203,7 +200,6 @@ def test_check_negative(scitype, mtype):
             # retrieve fixture for checking
             fixture_wrong_type = fixtures[wrong_mtype].get(i)
 
-            # todo: possibly remove this once all checks are defined
             check_is_defined = (mtype, scitype) in check_dict.keys()
 
             # check fixtures that exist against checks that exist
@@ -231,8 +227,6 @@ def test_mtype_infer(scitype, mtype, fixture_index):
     """
     # retrieve fixture for checking
     fixture = get_examples(mtype=mtype, as_scitype=scitype).get(fixture_index)
-
-    # todo: possibly remove this once all checks are defined
     check_is_defined = (mtype, scitype) in check_dict.keys()
 
     # check fixtures that exist against checks that exist
@@ -269,8 +263,6 @@ def test_scitype_infer(scitype, mtype, fixture_index):
 
     # retrieve fixture for checking
     fixture = get_examples(mtype=mtype, as_scitype=scitype).get(fixture_index)
-
-    # todo: possibly remove this once all checks are defined
     check_is_defined = (mtype, scitype) in check_dict.keys()
 
     # check fixtures that exist against checks that exist

aeon/forecasting/ardl.py

@@ -290,7 +290,6 @@ def check_param_validity(self, X):
                 )
         return inner_order, inner_auto_ardl
 
-    # todo: implement this, mandatory
     def _fit(self, y, X=None, fh=None):
         """Fit forecaster to training data.
 
@@ -468,8 +467,6 @@ def _update(self, y, X=None, update_params=True):
             mtype_last_seen = self._y_mtype_last_seen
             # refit with updated data, not only passed data
             self.fit(y=self._y, X=self._X, fh=self._fh)
-            # todo: should probably be self._fit, not self.fit
-            # but looping to self.fit for now to avoid interface break
             self._y_mtype_last_seen = mtype_last_seen
 
         # if update_params=False, and there are no components, do nothing

aeon/forecasting/base/_base.py

@@ -1421,18 +1421,13 @@ def _check_missing(metadata, obj_name):
         if X is not None and y is not None:
             if self.get_tag("X-y-must-have-same-index"):
                 # currently, check_equal_time_index only works for Series
-                # todo: fix this so the check is general, using get_time_index
                 if not self.get_tag("ignores-exogeneous-X") and X_scitype == "Series":
                     check_equal_time_index(X, y, mode="contains")
 
             if y_scitype != X_scitype:
                 raise TypeError("X and y must have the same scitype")
         # end compatibility checking X and y
 
-        # todo: add tests that :
-        #   y_inner_scitype are same as X_inner_scitype
-        #   y_inner_scitype always includes "less index" scitypes
-
         # convert X & y to supported inner type, if necessary
         #####################################################
 
@@ -1886,7 +1881,6 @@ def _update(self, y, X=None, update_params=True):
             _converter_store_y = self._converter_store_y
             # refit with updated data, not only passed data
             self.fit(y=self._y, X=self._X, fh=self._fh)
-            # todo: should probably be self._fit, not self.fit
             # but looping to self.fit for now to avoid interface break
             self._y_mtype_last_seen = mtype_last_seen
             self._converter_store_y = _converter_store_y
@@ -2125,7 +2119,6 @@ def _predict_var(self, fh=None, X=None, cov=False):
             )
 
         if implements_proba:
-            # todo: this works only univariate now, need to implement multivariate
             pred_var = self._predict_proba(fh=fh, X=X)
             pred_var = pd.DataFrame(pred_var)
 
@@ -2165,8 +2158,6 @@ def _predict_var(self, fh=None, X=None, cov=False):
 
         return pred_var
 
-    # todo: does not work properly for multivariate or hierarchical
-    #   still need to implement this - once interface is consolidated
     def _predict_proba(self, fh, X, marginal=True):
         """Compute/return fully probabilistic forecasts.
 

aeon/forecasting/base/_fh.py

@@ -686,24 +686,7 @@ def _to_relative(fh: ForecastingHorizon, cutoff=None) -> ForecastingHorizon:
             # time deltas
             absolute = _coerce_to_period(absolute, freq=fh.freq)
             cutoff = _coerce_to_period(cutoff, freq=fh.freq)
-
-        # TODO: Replace when we upgrade our lower pandas bound
-        #  to a version where this is fixed
-        # Compute relative values
-        # The following line circumvents the bug in pandas
-        # periods = pd.period_range(start="2021-01-01", periods=3, freq="2H")
-        # periods - periods[0]
-        # Out: Index([<0 * Hours>, <4 * Hours>, <8 * Hours>], dtype = 'object')
-        # [v - periods[0] for v in periods]
-        # Out: Index([<0 * Hours>, <2 * Hours>, <4 * Hours>], dtype='object')
-        # TODO: 0.17.0: Check if this comment below can be removed,
-        # so check if pandas has released the fix to PyPI:
-        # This bug was reported: https://github.com/pandas-dev/pandas/issues/45999
-        # and fixed: https://github.com/pandas-dev/pandas/pull/46006
-        # Most likely it will be released with pandas 1.5
-        # Once the bug is fixed the line should simply be:
-        # relative = absolute - cutoff
-        relative = pd.Index([date - cutoff for date in absolute])
+        relative = absolute - cutoff
 
         # Coerce durations (time deltas) into integer values for given frequency
         if isinstance(absolute, (pd.PeriodIndex, pd.DatetimeIndex)):

aeon/forecasting/compose/_pipeline.py

@@ -1461,8 +1461,6 @@ def _predict_var(self, fh=None, X=None, cov=False):
         y_pred = self.forecaster_y_.predict_var(fh=fh, X=X)
         return y_pred
 
-    # todo: does not work properly for multivariate or hierarchical
-    #   still need to implement this - once interface is consolidated
     def _predict_proba(self, fh, X, marginal=True):
         """Compute/return fully probabilistic forecasts.
 

aeon/forecasting/compose/_reduce.py

@@ -973,7 +973,6 @@ def _fit(self, y, X=None, fh=None):
         self : Estimator
             An fitted instance of self.
         """
-        # todo: logic for X below is broken. Escape X until fixed.
         if X is not None:
             X = None
 
@@ -1033,7 +1032,7 @@ def _predict_last_window(
         y_pred = pd.Series or pd.DataFrame
         """
         # Exogenous variables are not yet support for the dirrec strategy.
-        # todo: implement this. For now, we escape.
+
         if X is not None:
             X = None
 
@@ -2144,21 +2143,20 @@ def _fit(self, y, X=None, fh=None):
         Parameters
         ----------
         y : pd.DataFrame
-            mtype is pd.DataFrame, pd-multiindex, or pd_multiindex_hier
+            pd.DataFrame, pd-multiindex, or pd_multiindex_hier
             Time series to which to fit the forecaster.
         fh : guaranteed to be ForecastingHorizon or None, optional (default=None)
             The forecasting horizon with the steps ahead to to predict.
             Required (non-optional) here if self.get_tag("requires-fh-in-fit")==True
             Otherwise, if not passed in _fit, guaranteed to be passed in _predict
         X : pd.DataFrame optional (default=None)
-            mtype is pd.DataFrame, pd-multiindex, or pd_multiindex_hier
+            pd.DataFrame, pd-multiindex, or pd_multiindex_hier
             Exogeneous time series to fit to.
 
         Returns
         -------
         self : reference to self
         """
-        # todo: very similar to _fit_concurrent of DirectReductionForecaster - refactor?
         from aeon.transformations.impute import Imputer
         from aeon.transformations.lag import Lag
 
@@ -2211,7 +2209,7 @@ def _predict(self, X=None, fh=None):
             The forecasting horizon with the steps ahead to to predict.
             If not passed in _fit, guaranteed to be passed here
         X : pd.DataFrame, optional (default=None)
-            mtype is pd.DataFrame, pd-multiindex, or pd_multiindex_hier
+            pd.DataFrame, pd-multiindex, or pd_multiindex_hier
             Exogeneous time series for the forecast
 
         Returns
@@ -2378,7 +2376,7 @@ def get_test_params(cls, parameter_set="default"):
         params1 = {
             "estimator": est,
             "window_length": 3,
-            "pooling": "global",  # all internal mtypes are tested across scenarios
+            "pooling": "global",
         }
 
         return params1

aeon/forecasting/model_selection/tests/test_split.py

@@ -475,11 +475,6 @@ def test_split_by_fh(index_type, fh_type, is_relative, values):
     """Test temporal_train_test_split."""
     if fh_type == "timedelta":
         return None
-        # todo: ensure check_estimator works with pytest.skip like below
-        # pytest.skip(
-        #    "ForecastingHorizon with timedelta values "
-        #     "is currently experimental and not supported everywhere"
-        # )
     y = _make_series(20, index_type=index_type)
     cutoff = get_cutoff(y.iloc[:10], return_index=True)
     fh = _make_fh(cutoff, values, fh_type, is_relative)

aeon/forecasting/naive.py

@@ -120,7 +120,6 @@ def __init__(self, strategy="last", window_length=None, sp=1):
         self.window_length = window_length
 
         # Override tag for handling missing data
-        # todo: remove if GH1367 is fixed
         if self.strategy in ("last", "mean"):
             self.set_tags(**{"capability:missing_values": True})
 
@@ -335,7 +334,6 @@ def _predict(self, fh=None, X=None):
         # test_predict_time_index_in_sample_full[ForecastingPipeline-0-int-int-True]
         #   causes a pd.DataFrame to appear as y_pred, which upsets the next lines
         #   reasons are unclear, this is coming from the _BaseWindowForecaster
-        # todo: investigate this
         if isinstance(y_pred, pd.DataFrame):
             y_pred = y_pred.iloc[:, 0]
 

aeon/forecasting/reconcile.py

@@ -5,7 +5,6 @@
     "ciaran-g",
 ]
 
-# todo: top down historical proportions? -> new _get_g_matrix_prop(self)
 
 from warnings import warn
 

aeon/forecasting/tests/test_all_forecasters.py

@@ -148,7 +148,6 @@ def test_get_fitted_params(self, estimator_instance, scenario):
         except NotImplementedError:
             pass
 
-    # todo: should these not be checked in test_all_estimators?
     def test_raises_not_fitted_error(self, estimator_instance):
         """Test that calling post-fit methods before fit raises error."""
         # We here check extra method of the forecaster API: update and update_predict.
@@ -177,14 +176,12 @@ def test_y_multivariate_raises_error(self, estimator_instance):
             # and "univariate" automatically vectorizes, behaves multivariate
             pass
 
-    # todo: should these not be "negative scenarios", tested in test_all_estimators?
     @pytest.mark.parametrize("y", INVALID_y_INPUT_TYPES)
     def test_y_invalid_type_raises_error(self, estimator_instance, y):
         """Test that invalid y input types raise error."""
         with pytest.raises(TypeError, match=r"type"):
             estimator_instance.fit(y, fh=FH0)
 
-    # todo: should these not be "negative scenarios", tested in test_all_estimators?
     @pytest.mark.parametrize("X", INVALID_X_INPUT_TYPES)
     def test_X_invalid_type_raises_error(self, estimator_instance, n_columns, X):
         """Test that invalid X input types raise error."""
@@ -208,11 +205,6 @@ def test_predict_residuals(
         if fh_type == "timedelta":
             # workaround to ensure check_estimator without breaking e.g. debugging
             return None
-            # todo: ensure check_estimator works with pytest.skip like below
-            # pytest.skip(
-            #    "ForecastingHorizon with timedelta values "
-            #     "is currently experimental and not supported everywhere"
-            # )
         y_train = _make_series(
             n_columns=n_columns, index_type=index_type, n_timepoints=50
         )
@@ -244,11 +236,6 @@ def test_predict_time_index_with_X(
         index_type, fh_type, is_relative = index_fh_comb
         if fh_type == "timedelta":
             return None
-            # todo: ensure check_estimator works with pytest.skip like below
-            # pytest.skip(
-            #    "ForecastingHorizon with timedelta values "
-            #     "is currently experimental and not supported everywhere"
-            # )
         z, X = make_forecasting_problem(index_type=index_type, make_X=True)
 
         # Some estimators may not support all time index types and fh types, hence we
@@ -278,11 +265,6 @@ def test_predict_time_index_in_sample_full(
         index_type, fh_type, is_relative = index_fh_comb
         if fh_type == "timedelta":
             return None
-            # todo: ensure check_estimator works with pytest.skip like below
-            # pytest.skip(
-            #    "ForecastingHorizon with timedelta values "
-            #     "is currently experimental and not supported everywhere"
-            # )
         y_train = _make_series(n_columns=n_columns, index_type=index_type)
         cutoff = get_cutoff(y_train, return_index=True)
         steps = -np.arange(len(y_train))

aeon/forecasting/theta.py

@@ -226,9 +226,6 @@ def _predict_quantiles(self, fh, X=None, alpha=None):
         # we assume normal additive noise with sem variance
         for a in alpha:
             pred_quantiles[("Quantiles", a)] = y_pred + norm.ppf(a) * sem
-        # todo: should this not increase with the horizon?
-        # i.e., sth like norm.ppf(a) * sem * fh.to_absolute(cutoff) ?
-        # I've just refactored this so will leave it for now
 
         return pred_quantiles
 

aeon/performance_metrics/forecasting/_classes.py

@@ -430,7 +430,7 @@ def _check_consistent_input(self, y_true, y_pred, multioutput, multilevel):
             else:
                 y_pred_orig.columns = y_true.columns
         # check multioutput arg
-        # todo: add this back when variance_weighted is supported
+        # add this back when variance_weighted is supported
         # ("raw_values", "uniform_average", "variance_weighted")
         allowed_multioutput_str = ("raw_values", "uniform_average")
 

aeon/performance_metrics/forecasting/probabilistic/_classes.py

@@ -8,8 +8,6 @@
 from aeon.datatypes import check_is_scitype, convert
 from aeon.performance_metrics.forecasting._classes import BaseForecastingErrorMetric
 
-# TODO: Rework tests now
-
 
 class _BaseProbaForecastingErrorMetric(BaseForecastingErrorMetric):
     """Base class for probabilistic forecasting error metrics in aeon.
@@ -419,8 +417,6 @@ def _evaluate_by_index(self, y_true, y_pred, multioutput, **kwargs):
             # if alpha was provided, check whether  they are predicted
             #   if not all alpha are observed, raise a ValueError
             if not np.isin(alpha, y_pred_alphas).all():
-                # todo: make error msg more informative
-                #   which alphas are missing
                 msg = "not all quantile values in alpha are available in y_pred"
                 raise ValueError(msg)
             else:

aeon/tests/test_all_estimators.py

@@ -629,8 +629,6 @@ def _check_None_str_or_list_of_str(obj, var_name="obj"):
                 raise ValueError(msg)
         return obj
 
-    # todo: surely there is a pytest method that can be called instead of this?
-    #   find and replace if it exists
     @staticmethod
     def _get_pytest_mark_args(fun):
         """Get args from pytest mark annotation of function.
@@ -1319,7 +1317,6 @@ def test_dl_constructor_initializes_deeply(self, estimator_class):
             if vars(estimator._network).get(key) is not None:
                 assert vars(estimator._network)[key] == value
 
-    # todo: this needs to be diagnosed and fixed - temporary skip
     @pytest.mark.skip(reason="hangs on mac and unix remote tests")
     def test_multiprocessing_idempotent(
         self, estimator_instance, scenario, method_nsc_arraylike

aeon/tests/test_softdeps.py

@@ -39,8 +39,6 @@
 # for the reason that they are composites which have soft dependencies in examples
 # but no soft dependencies themselves, so it's temporarily fine to raise this
 # e.g., forecasting pipeline with an ARIMA estimator
-# todo: long-term all example parameter settings should be soft dependency free
-# strings of class names to avoid the imports
 EXCEPTED_FROM_NO_DEP_CHECK = []
 
 

aeon/transformations/boxcox.py

@@ -13,7 +13,7 @@
 
 
 # copy-pasted from scipy 1.7.3 since it moved in 1.8.0 and broke this estimator
-# todo: find a suitable replacement
+# find a suitable replacement
 def _calc_uniform_order_statistic_medians(n):
     """Approximations of uniform order statistic medians.