[ENH] Remove make_series part II (#2310)

* anomaly tests

* anomaly tests

* stomp and kmeansad

* pyod adaptor

* pyod adaptor

* pyod adaptor

* iforest

* seed

* split functions, use check_random

* test clean up

* check_series

* fix soft-dtw cost matrix test

* check_series

* pairwise

* viz

* remove make_series

* Update test_forecasting_plotting.py

* Update test_segmentation_plotting.py

* fixing distance tests

* fixed distance tests

* fix shape-dtw docs

---------

Co-authored-by: Sebastian Schmidl <CodeLionX@users.noreply.github.com>
Co-authored-by: chrisholder <chrisholder987@hotmail.com>
Co-authored-by: Matthew Middlehurst <pfm15hbu@gmail.com>

Author: 4 people

aeon/clustering/averaging/tests/test_dba.py

@@ -10,8 +10,11 @@
     petitjean_barycenter_average,
     subgradient_barycenter_average,
 )
-from aeon.testing.data_generation import make_example_3d_numpy
-from aeon.testing.data_generation._legacy import make_series
+from aeon.testing.data_generation import (
+    make_example_1d_numpy,
+    make_example_2d_numpy_series,
+    make_example_3d_numpy,
+)
 
 expected_petitjean_dba_univariate = np.array(
     [
@@ -231,10 +234,8 @@ def test_elastic_dba_variations(distance):
         "medoids",
         "random",
         (
-            make_series(10, 1, return_numpy=True, random_state=1),
-            make_series(
-                n_timepoints=4, n_columns=10, return_numpy=True, random_state=1
-            ),
+            make_example_1d_numpy(10, random_state=1),
+            make_example_2d_numpy_series(n_timepoints=10, n_channels=4, random_state=1),
         ),
     ],
 )
@@ -303,7 +304,7 @@ def test_incorrect_input():
             "init_barycenter shape is invalid. Expected (1, 10) but " "got (1, 9)"
         ),
     ):
-        elastic_barycenter_average(X, init_barycenter=make_series(9, return_numpy=True))
+        elastic_barycenter_average(X, init_barycenter=make_example_1d_numpy(9))
 
     # Test invalid berycenter method
     with pytest.raises(

aeon/distances/elastic/_msm.py

@@ -247,7 +247,8 @@ def _msm_independent_cost_matrix(
     y_size = y.shape[1]
     cost_matrix = np.zeros((x_size, y_size))
     distance = 0
-    for i in range(x.shape[0]):
+    min_instances = min(x.shape[0], y.shape[0])
+    for i in range(min_instances):
         curr_cost_matrix = _independent_cost_matrix(x[i], y[i], bounding_matrix, c)
         cost_matrix = np.add(cost_matrix, curr_cost_matrix)
         distance += curr_cost_matrix[-1, -1]

aeon/distances/elastic/_shape_dtw.py

@@ -60,7 +60,7 @@ def _identity_descriptor(x: np.ndarray) -> np.ndarray:
 
 @njit(cache=True, fastmath=True)
 def _transform_subsequences(
-    x: np.ndarray, descriptor: str = "identity", reach: int = 30
+    x: np.ndarray, descriptor: str = "identity", reach: int = 15
 ) -> np.ndarray:
     """Decompose the series into sub-sequences.
 
@@ -80,7 +80,7 @@ def _transform_subsequences(
         For now no other descriptors are implemented.
 
         If not specified then identity is used.
-    reach : int, default=30.
+    reach : int, default=15.
         Length of the sub-sequences.
 
     Returns
@@ -123,7 +123,7 @@ def shape_dtw_distance(
     y: np.ndarray,
     window: Optional[float] = None,
     descriptor: str = "identity",
-    reach: int = 30,
+    reach: int = 15,
     itakura_max_slope: Optional[float] = None,
     transformation_precomputed: bool = False,
     transformed_x: Optional[np.ndarray] = None,
@@ -156,7 +156,7 @@ def shape_dtw_distance(
         For now no other descriptors are implemented.
 
         If not specified then identity is used.
-    reach : int, default=30.
+    reach : int, default=15.
         Length of the sub-sequences to consider.
     itakura_max_slope : float, default=None
         Maximum slope as a proportion of the number of time points used to create
@@ -237,7 +237,7 @@ def _shape_dtw_distance(
     y: np.ndarray,
     bounding_matrix: np.ndarray,
     descriptor: str = "identity",
-    reach: int = 30,
+    reach: int = 15,
     transformation_precomputed: bool = False,
     transformed_x: Optional[np.ndarray] = None,
     transformed_y: Optional[np.ndarray] = None,
@@ -297,7 +297,7 @@ def shape_dtw_cost_matrix(
     y: np.ndarray,
     window: Optional[float] = None,
     descriptor: str = "identity",
-    reach: int = 30,
+    reach: int = 15,
     itakura_max_slope: Optional[float] = None,
     transformation_precomputed: bool = False,
     transformed_x: Optional[np.ndarray] = None,
@@ -327,7 +327,7 @@ def shape_dtw_cost_matrix(
         For now no other descriptors are implemented.
 
         If not specified then identity is used.
-    reach : int, default=30.
+    reach : int, default=15.
         Length of the sub-sequences.
     itakura_max_slope : float, default=None
         Maximum slope as a proportion of the number of time points used to create
@@ -398,7 +398,7 @@ def _shape_dtw_cost_matrix(
     y: np.ndarray,
     bounding_matrix: np.ndarray,
     descriptor: str = "identity",
-    reach: int = 30,
+    reach: int = 15,
     transformation_precomputed: bool = False,
     transformed_x: Optional[np.ndarray] = None,
     transformed_y: Optional[np.ndarray] = None,
@@ -430,7 +430,7 @@ def shape_dtw_alignment_path(
     y: np.ndarray,
     window: Optional[float] = None,
     descriptor: str = "identity",
-    reach: int = 30,
+    reach: int = 15,
     itakura_max_slope: Optional[float] = None,
     transformation_precomputed: bool = False,
     transformed_x: Optional[np.ndarray] = None,
@@ -460,7 +460,7 @@ def shape_dtw_alignment_path(
         For now no other descriptors are implemented.
 
         If not specified then identity is used.
-    reach : int, default=30.
+    reach : int, default=15.
         Length of the sub-sequences.
     itakura_max_slope : float, default=None
         Maximum slope as a proportion of the number of time points used to create
@@ -520,7 +520,7 @@ def shape_dtw_pairwise_distance(
     y: Optional[Union[np.ndarray, list[np.ndarray]]] = None,
     window: Optional[float] = None,
     descriptor: str = "identity",
-    reach: int = 30,
+    reach: int = 15,
     itakura_max_slope: Optional[float] = None,
     transformation_precomputed: bool = False,
     transformed_x: Optional[np.ndarray] = None,
@@ -552,7 +552,7 @@ def shape_dtw_pairwise_distance(
         For now no other descriptors are implemented.
 
         If not specified then identity is used.
-    reach : int, default=30.
+    reach : int, default=15.
         Length of the sub-sequences.
     itakura_max_slope : float, default=None
         Maximum slope as a proportion of the number of time points used to create

aeon/distances/elastic/tests/test_alignment_path.py

@@ -10,7 +10,10 @@
     DISTANCES_DICT,
     SINGLE_POINT_NOT_SUPPORTED_DISTANCES,
 )
-from aeon.testing.data_generation._legacy import make_series
+from aeon.testing.data_generation import (
+    make_example_1d_numpy,
+    make_example_2d_numpy_series,
+)
 
 
 def _validate_alignment_path_result(
@@ -57,26 +60,26 @@ def test_alignment_path(dist):
     # ================== Test equal length ==================
     # Test univariate of shape (n_timepoints,)
     _validate_alignment_path_result(
-        make_series(10, return_numpy=True, random_state=1),
-        make_series(10, return_numpy=True, random_state=2),
+        make_example_1d_numpy(10, random_state=1),
+        make_example_1d_numpy(10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["alignment_path"],
     )
 
     # Test univariate of shape (1, n_timepoints)
     _validate_alignment_path_result(
-        make_series(10, 1, return_numpy=True, random_state=1),
-        make_series(10, 1, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(10, 1, random_state=1),
+        make_example_2d_numpy_series(10, 1, random_state=1),
         dist["name"],
         dist["distance"],
         dist["alignment_path"],
     )
 
     # Test multivariate of shape (n_channels, n_timepoints)
     _validate_alignment_path_result(
-        make_series(10, 10, return_numpy=True, random_state=1),
-        make_series(10, 10, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(10, 10, random_state=1),
+        make_example_2d_numpy_series(10, 10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["alignment_path"],
@@ -85,26 +88,26 @@ def test_alignment_path(dist):
     # ================== Test unequal length ==================
     # Test univariate unequal length of shape (n_timepoints,)
     _validate_alignment_path_result(
-        make_series(5, return_numpy=True, random_state=1),
-        make_series(10, return_numpy=True, random_state=2),
+        make_example_1d_numpy(5, random_state=1),
+        make_example_1d_numpy(10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["alignment_path"],
     )
 
     # Test univariate unequal length of shape (1, n_timepoints)
     _validate_alignment_path_result(
-        make_series(5, 1, return_numpy=True, random_state=1),
-        make_series(10, 1, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(5, 1, random_state=1),
+        make_example_2d_numpy_series(10, 1, random_state=2),
         dist["name"],
         dist["distance"],
         dist["alignment_path"],
     )
 
     # Test multivariate unequal length of shape (n_channels, n_timepoints)
     _validate_alignment_path_result(
-        make_series(5, 10, return_numpy=True, random_state=1),
-        make_series(10, 10, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(5, 10, random_state=1),
+        make_example_2d_numpy_series(10, 10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["alignment_path"],

aeon/distances/elastic/tests/test_cost_matrix.py

@@ -10,7 +10,10 @@
     DISTANCES_DICT,
     SINGLE_POINT_NOT_SUPPORTED_DISTANCES,
 )
-from aeon.testing.data_generation._legacy import make_series
+from aeon.testing.data_generation import (
+    make_example_1d_numpy,
+    make_example_2d_numpy_series,
+)
 
 
 def _validate_cost_matrix_result(
@@ -61,6 +64,8 @@ def _validate_cost_matrix_result(
                 cost_matrix_result[-1, -1] / max(x.shape[-1], y.shape[-1])
             )
             assert_almost_equal(curr_distance, distance_result)
+    elif name == "soft_dtw":
+        assert_almost_equal(abs(cost_matrix_result[-1, -1]), distance_result)
     else:
         assert_almost_equal(cost_matrix_result[-1, -1], distance_result)
 
@@ -88,26 +93,26 @@ def test_cost_matrix(dist):
     # ================== Test equal length ==================
     # Test univariate of shape (n_timepoints,)
     _validate_cost_matrix_result(
-        make_series(10, return_numpy=True, random_state=1),
-        make_series(10, return_numpy=True, random_state=2),
+        make_example_1d_numpy(10, random_state=1),
+        make_example_1d_numpy(10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["cost_matrix"],
     )
 
     # Test univariate of shape (1, n_timepoints)
     _validate_cost_matrix_result(
-        make_series(10, 1, return_numpy=True, random_state=1),
-        make_series(10, 1, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(10, 1, random_state=1),
+        make_example_2d_numpy_series(10, 1, random_state=2),
         dist["name"],
         dist["distance"],
         dist["cost_matrix"],
     )
 
     # Test multivariate of shape (n_channels, n_timepoints)
     _validate_cost_matrix_result(
-        make_series(10, 10, return_numpy=True, random_state=1),
-        make_series(10, 10, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(10, 10, random_state=1),
+        make_example_2d_numpy_series(10, 10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["cost_matrix"],
@@ -116,26 +121,26 @@ def test_cost_matrix(dist):
     # ================== Test unequal length ==================
     # Test univariate unequal length of shape (n_timepoints,)
     _validate_cost_matrix_result(
-        make_series(5, return_numpy=True, random_state=1),
-        make_series(10, return_numpy=True, random_state=2),
+        make_example_1d_numpy(5, random_state=1),
+        make_example_1d_numpy(10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["cost_matrix"],
     )
 
     # Test univariate unequal length of shape (1, n_timepoints)
     _validate_cost_matrix_result(
-        make_series(5, 1, return_numpy=True, random_state=1),
-        make_series(10, 1, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(5, 1, random_state=1),
+        make_example_2d_numpy_series(10, 1, random_state=2),
         dist["name"],
         dist["distance"],
         dist["cost_matrix"],
     )
 
     # Test multivariate unequal length of shape (n_channels, n_timepoints)
     _validate_cost_matrix_result(
-        make_series(5, 10, return_numpy=True, random_state=1),
-        make_series(10, 10, return_numpy=True, random_state=2),
+        make_example_2d_numpy_series(5, 10, random_state=1),
+        make_example_2d_numpy_series(10, 10, random_state=2),
         dist["name"],
         dist["distance"],
         dist["cost_matrix"],