[BUG] Fix redcomets bug when using only one sample (#2952)

* Fix

* fix exemples

* Empty commit for CI

* Rework examples and add test

* Try fixing example indentation error

* Fix examples

---------

Co-authored-by: baraline <10759117+baraline@users.noreply.github.com>
Co-authored-by: Patrick Schäfer <patrick.schaefer@hu-berlin.de>

Author: baraline

aeon/classification/dictionary_based/_redcomets.py

@@ -150,7 +150,7 @@ def _fit(self, X, y):
                 self.sfa_clfs,
                 self.sax_transforms,
                 self.sax_clfs,
-            ) = self._build_univariate_ensemble(np.squeeze(X), y)
+            ) = self._build_univariate_ensemble(np.squeeze(X, 1), y)
         else:  # Multivariate
 
             if self.variant in [1, 2, 3]:  # Concatenate
@@ -204,7 +204,7 @@ def _build_univariate_ensemble(self, X, y):
 
         from imblearn.over_sampling import SMOTE, RandomOverSampler
 
-        X = Normalizer().fit_transform(X).squeeze()
+        X = Normalizer().fit_transform(X).squeeze(1)
 
         if self.variant in [1, 2, 3]:
             perc_length = self.perc_length / self._n_channels
@@ -391,7 +391,7 @@ def _predict_proba(self, X) -> np.ndarray:
             Predicted probabilities using the ordering in ``classes_``.
         """
         if X.shape[1] == 1:  # Univariate
-            return self._predict_proba_unvivariate(np.squeeze(X))
+            return self._predict_proba_unvivariate(np.squeeze(X, 1))
         else:  # Multivariate
             if self.variant in [1, 2, 3]:  # Concatenate
                 X_concat = X.reshape(*X.shape[:-2], -1)
@@ -414,7 +414,7 @@ def _predict_proba_unvivariate(self, X) -> np.ndarray:
             2D np.ndarray of shape (n_cases, n_classes_)
             Predicted probabilities using the ordering in ``classes_``.
         """
-        X = Normalizer().fit_transform(X).squeeze()
+        X = Normalizer().fit_transform(X).squeeze(1)
 
         pred_mat = np.zeros((X.shape[0], self.n_classes_))
 
@@ -588,7 +588,7 @@ def _parallel_sax(self, sax_transforms, X):
         """
 
         def _sax_wrapper(sax):
-            return np.squeeze(sax.fit_transform(X))
+            return np.squeeze(sax.fit_transform(X), 1)
 
         sax_parallel_res = Parallel(n_jobs=self._n_jobs, backend=self.parallel_backend)(
             delayed(_sax_wrapper)(sax) for sax in sax_transforms

aeon/distances/mindist/_dft_sfa.py

@@ -45,20 +45,24 @@ def mindist_dft_sfa_distance(
     >>> import numpy as np
     >>> from aeon.distances import mindist_dft_sfa_distance
     >>> from aeon.transformations.collection.dictionary_based import SFAWhole
-    >>> x = np.array([[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]])
-    >>> y = np.array([[11, 12, 13, 14, 15, 16, 17, 18, 19, 20]])
+    >>> x = np.random.rand(1,1,10) # (n_cases, n_channels, n_timepoints)
+    >>> y = np.random.rand(1,1,10)
     >>> transform = SFAWhole(
     ...    word_length=8,
     ...    alphabet_size=8,
     ...    norm=True,
     ... )
     >>> x_sfa, _ = transform.fit_transform(x)
     >>> _, y_dft = transform.transform(y)
-    >>> dist = mindist_dft_sfa_distance(y_dft, x_sfa, transform.breakpoints)
+    >>> for i in range(x.shape[0]):
+    ...    dist = mindist_dft_sfa_distance(y_dft[0], x_sfa[0], transform.breakpoints)
     """
     if x_dft.ndim == 1 and y_sfa.ndim == 1:
         return _univariate_dft_sfa_distance(x_dft, y_sfa, breakpoints)
-    raise ValueError("x and y must be 1D")
+    raise ValueError(
+        f"x and y must be 1D, but got x of shape {x_dft.shape} and y of shape"
+        f"{y_sfa.shape}"
+    )
 
 
 @njit(cache=True, fastmath=True)

aeon/distances/mindist/_sfa.py

@@ -45,20 +45,23 @@ def mindist_sfa_distance(
     >>> import numpy as np
     >>> from aeon.distances import mindist_sfa_distance
     >>> from aeon.transformations.collection.dictionary_based import SFAWhole
-    >>> x = np.array([[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]])
-    >>> y = np.array([[11, 12, 13, 14, 15, 16, 17, 18, 19, 20]])
+    >>> x = np.random.rand(1,1,10) # (n_cases, n_channels, n_timepoints)
+    >>> y = np.random.rand(1,1,10)
     >>> transform = SFAWhole(
     ...    word_length=8,
     ...    alphabet_size=8,
     ...    norm=True
     ... )
     >>> x_sfa, _ = transform.fit_transform(x)
     >>> y_sfa, _ = transform.transform(y)
-    >>> dist = mindist_sfa_distance(x_sfa, y_sfa, transform.breakpoints)
+    >>> for i in range(x.shape[0]):
+    ...    dist = mindist_sfa_distance(x_sfa[i], y_sfa[i], transform.breakpoints)
     """
     if x.ndim == 1 and y.ndim == 1:
         return _univariate_sfa_distance(x, y, breakpoints)
-    raise ValueError("x and y must be 1D")
+    raise ValueError(
+        f"x and y must be 1D, but got x of shape {x.shape} and y of shape {y.shape}"
+    )
 
 
 @njit(cache=True, fastmath=True)

aeon/distances/tests/test_symbolic_mindist.py

@@ -8,6 +8,7 @@
 from aeon.distances.mindist._paa_sax import mindist_paa_sax_distance
 from aeon.distances.mindist._sax import mindist_sax_distance
 from aeon.distances.mindist._sfa import mindist_sfa_distance
+from aeon.testing.data_generation import make_example_3d_numpy
 from aeon.transformations.collection.dictionary_based import SAX, SFA, SFAFast, SFAWhole
 
 
@@ -49,6 +50,18 @@ def test_sax_mindist():
         assert mindist_paa_sax <= ed
 
 
+def test_single_sample():
+    """Test the SFA Min-Distance function."""
+    x, _ = make_example_3d_numpy(n_cases=1, n_channels=1, n_timepoints=10)
+    y = x + 10
+    transform = SFAWhole(word_length=8, alphabet_size=8, norm=True)
+    x_sfa, _ = transform.fit_transform(x)
+    _, y_dft = transform.transform(y)
+    for i in range(len(x_sfa)):
+        dist = mindist_dft_sfa_distance(y_dft[i], x_sfa[i], transform.breakpoints)
+        assert dist == 0
+
+
 def test_sfa_mindist():
     """Test the SFA Min-Distance function."""
     n_segments = 16

aeon/transformations/collection/dictionary_based/_sfa_fast.py

@@ -729,8 +729,7 @@ def transform_words(self, X):
             self.alphabet_size,
             self.breakpoints,
         )
-
-        return words.squeeze(), dfts.squeeze()
+        return words.squeeze(1), dfts.squeeze(1)
 
     @classmethod
     def _get_test_params(cls, parameter_set="default"):