[ENH] Resolve failing check_fit_deterministic

aeon/classification/shapelet_based/_rsast.py

@@ -7,6 +7,7 @@
 __all__ = ["RSASTClassifier"]
 
 import numpy as np
+from deprecated.sphinx import deprecated
 from sklearn.linear_model import RidgeClassifierCV
 from sklearn.pipeline import make_pipeline
 
@@ -30,8 +31,10 @@ class RSASTClassifier(BaseClassifier):
     "None"=Extract randomly any length from the TS
     nb_inst_per_class : int default = 10
         the number of reference time series to select per class
-    seed : int, default = None
+    random_state : int, default = None
         the seed of the random generator
+    seed : int, default= None
+        Deprecated and will be removed in v1.2. Use `random_state` instead.
     estimator : sklearn compatible classifier, default = None
         if None, a RidgeClassifierCV(alphas=np.logspace(-3, 3, 10)) is used.
     n_jobs : int, default -1
@@ -63,21 +66,49 @@ class RSASTClassifier(BaseClassifier):
         "python_dependencies": "statsmodels",
     }
 
+    # TODO: remove 'seed' in v1.2
+    @deprecated(
+        version="1.1",
+        reason="The 'seed' parameter will be removed in v1.2.",
+        category=FutureWarning,
+    )
     def __init__(
         self,
         n_random_points=10,
         len_method="both",
         nb_inst_per_class=10,
-        seed=None,
+        random_state=None,
         classifier=None,
         n_jobs=1,
+        seed=None,
     ):
         super().__init__()
         self.n_random_points = n_random_points
         self.len_method = len_method
         self.nb_inst_per_class = nb_inst_per_class
         self.n_jobs = n_jobs
+        # Store the seed parameter (required for sklearn compatibility)
         self.seed = seed
+
+        # Handle deprecated seed parameter
+        if seed is not None:
+            import warnings
+
+            warnings.warn(
+                "The 'seed' parameter is deprecated and will be removed in v1.2. "
+                "Use 'random_state' instead.",
+                FutureWarning,
+                stacklevel=2,
+            )
+            if random_state is None:
+                random_state = seed
+            else:
+                raise ValueError(
+                    "Cannot specify both 'seed' and 'random_state'. "
+                    "Use 'random_state' only."
+                )
+
+        self.random_state = random_state
         self.classifier = classifier
 
     def _fit(self, X, y):
@@ -100,7 +131,7 @@ def _fit(self, X, y):
             self.n_random_points,
             self.len_method,
             self.nb_inst_per_class,
-            self.seed,
+            self.random_state,
             self.n_jobs,
         )
 
@@ -110,7 +141,7 @@ def _fit(self, X, y):
                 if self.classifier is None
                 else self.classifier
             ),
-            self.seed,
+            self.random_state,
         )
 
         self._pipeline = make_pipeline(self._transformer, self._classifier)

aeon/classification/shapelet_based/_sast.py

@@ -11,6 +11,7 @@
 from operator import itemgetter
 
 import numpy as np
+from deprecated.sphinx import deprecated
 from sklearn.linear_model import RidgeClassifierCV
 from sklearn.pipeline import make_pipeline
 
@@ -32,8 +33,10 @@ class SASTClassifier(BaseClassifier):
         the stride used when generating subsquences
     nb_inst_per_class : int default = 1
         the number of reference time series to select per class
-    seed : int, default = None
+    random_state : int, default = None
         the seed of the random generator
+    seed : int, default=None
+        Deprecated and will be removed in v1.2. Use `random_state` instead.
     estimator : sklearn compatible classifier, default = None
         if None, a RidgeClassifierCV(alphas=np.logspace(-3, 3, 10)) is used.
     n_jobs : int, default -1
@@ -66,22 +69,50 @@ class SASTClassifier(BaseClassifier):
         "algorithm_type": "shapelet",
     }
 
+    # TODO: remove 'seed' in v1.2
+    @deprecated(
+        version="1.1",
+        reason="The 'seed' parameter will be removed in v1.2.",
+        category=FutureWarning,
+    )
     def __init__(
         self,
         length_list=None,
         stride: int = 1,
         nb_inst_per_class: int = 1,
-        seed: Optional[int] = None,
+        random_state: Optional[int] = None,
         classifier=None,
         n_jobs: int = 1,
+        seed=None,
     ) -> None:
         super().__init__()
         self.length_list = length_list
         self.stride = stride
         self.nb_inst_per_class = nb_inst_per_class
         self.n_jobs = n_jobs
+
+        # Store the seed parameter (required for sklearn compatibility)
         self.seed = seed
 
+        # Handle deprecated seed parameter
+        if seed is not None:
+            import warnings
+
+            warnings.warn(
+                "The 'seed' parameter is deprecated and will be removed in v1.2. "
+                "Use 'random_state' instead.",
+                FutureWarning,
+                stacklevel=2,
+            )
+            if random_state is None:
+                random_state = seed
+            else:
+                raise ValueError(
+                    "Cannot specify both 'seed' and 'random_state'. "
+                    "Use 'random_state' only."
+                )
+
+        self.random_state = random_state
         self.classifier = classifier
 
     def _fit(self, X, y):
@@ -104,7 +135,7 @@ def _fit(self, X, y):
             self.length_list,
             self.stride,
             self.nb_inst_per_class,
-            self.seed,
+            self.random_state,
             self.n_jobs,
         )
 
@@ -114,7 +145,7 @@ def _fit(self, X, y):
                 if self.classifier is None
                 else self.classifier
             ),
-            self.seed,
+            self.random_state,
         )
 
         self._pipeline = make_pipeline(self._transformer, self._classifier)

aeon/testing/testing_config.py

@@ -46,11 +46,6 @@
         "check_persistence_via_pickle",
         "check_save_estimators_to_file",
     ],
-    # needs investigation
-    "SASTClassifier": ["check_fit_deterministic"],
-    "RSASTClassifier": ["check_fit_deterministic"],
-    "SAST": ["check_fit_deterministic"],
-    "RSAST": ["check_fit_deterministic"],
     "MatrixProfile": ["check_fit_deterministic", "check_persistence_via_pickle"],
     "LeftSTAMPi": ["check_anomaly_detector_output"],
     # missed in legacy testing, changes state in predict/transform

aeon/transformations/collection/shapelet_based/_rsast.py

@@ -4,6 +4,7 @@
 
 import numpy as np
 import pandas as pd
+from deprecated.sphinx import deprecated
 from numba import get_num_threads, njit, prange, set_num_threads
 
 from aeon.transformations.collection import BaseCollectionTransformer
@@ -64,8 +65,10 @@ class RSAST(BaseCollectionTransformer):
 
     nb_inst_per_class : int default = 10
         the number of reference time series to select per class
-    seed : int, default = None
+    random_state : int, default = None
         the seed of the random generator
+    seed : int, default= None
+        Deprecated and will be removed in v1.2. Use `random_state` instead.
     n_jobs : int, default -1
         Number of threads to use for the transform.
 
@@ -98,26 +101,53 @@ class RSAST(BaseCollectionTransformer):
         "python_dependencies": "statsmodels",
     }
 
+    # TODO: remove 'seed' in v1.2
+    @deprecated(
+        version="1.1",
+        reason="The 'seed' parameter will be removed in v1.2.",
+        category=FutureWarning,
+    )
     def __init__(
         self,
         n_random_points: int = 10,
         len_method: str = "both",
         nb_inst_per_class: int = 10,
-        seed: Optional[int] = None,
+        random_state: Optional[int] = None,
         n_jobs: int = 1,  # Parllel Processing
+        seed=None,
     ):
         self.n_random_points = n_random_points
         self.len_method = len_method
         self.nb_inst_per_class = nb_inst_per_class
         self.n_jobs = n_jobs
-        self.seed = seed
         self._kernels = None  # z-normalized subsequences
         self._cand_length_list = {}
         self._kernel_orig = []
         self._start_points = []
         self._classes = []
         self._source_series = []  # To store the index of the original time series
         self._kernels_generators = {}  # Reference time series
+
+        # Handle deprecated seed parameter
+        # Store the seed parameter (required for sklearn compatibility)
+        self.seed = seed
+        if seed is not None:
+            import warnings
+
+            warnings.warn(
+                "The 'seed' parameter is deprecated and will be removed in v1.2. "
+                "Use 'random_state' instead.",
+                FutureWarning,
+                stacklevel=2,
+            )
+            if random_state is None:
+                random_state = seed
+            else:
+                raise ValueError(
+                    "Cannot specify both 'seed' and 'random_state'. "
+                    "Use 'random_state' only."
+                )
+        self.random_state = random_state
         super().__init__()
 
     def _fit(self, X: np.ndarray, y: Union[np.ndarray, list]) -> "RSAST":
@@ -144,9 +174,9 @@ def _fit(self, X: np.ndarray, y: Union[np.ndarray, list]) -> "RSAST":
         X_ = np.reshape(X, (X.shape[0], X.shape[-1]))
 
         self._random_state = (
-            np.random.RandomState(self.seed)
-            if not isinstance(self.seed, np.random.RandomState)
-            else self.seed
+            np.random.RandomState(self.random_state)
+            if not isinstance(self.random_state, np.random.RandomState)
+            else self.random_state
         )
 
         classes = np.unique(y)

aeon/transformations/collection/shapelet_based/_sast.py

@@ -3,6 +3,7 @@
 from typing import Optional, Union
 
 import numpy as np
+from deprecated.sphinx import deprecated
 from numba import get_num_threads, njit, prange, set_num_threads
 
 from aeon.transformations.collection import BaseCollectionTransformer
@@ -56,8 +57,10 @@ class SAST(BaseCollectionTransformer):
         the stride used when generating subsequences
     nb_inst_per_class : int, default = 1
         the number of reference time series to select per class
-    seed : int, default = None
+    random_state : int, default = None
         the seed of the random generator
+    seed : int, default=None
+        Deprecated and will be removed in v1.2. Use `random_state` instead.
     n_jobs : int, default -1
         Number of threads to use for the transform.
         The available CPU count is used if this value is less than 1
@@ -92,27 +95,57 @@ class SAST(BaseCollectionTransformer):
         "algorithm_type": "shapelet",
     }
 
+    # TODO: remove 'seed' in v1.2
+    @deprecated(
+        version="1.1",
+        reason="The 'seed' parameter will be removed in v1.2.",
+        category=FutureWarning,
+    )
     def __init__(
         self,
         lengths: Optional[np.ndarray] = None,
         stride: int = 1,
         nb_inst_per_class: int = 1,
-        seed: Optional[int] = None,
-        n_jobs: int = 1,  # Parallel processing
+        random_state: Optional[int] = None,
+        n_jobs: int = 1,
+        seed=None,
     ):
+
         super().__init__()
         self.lengths = lengths
         self.stride = stride
         self.nb_inst_per_class = nb_inst_per_class
-        self._kernels = None  # z-normalized subsequences
-        self._kernel_orig = None  # non z-normalized subsequences
-        self._start_points = []  # To store the start positions
-        self._classes = []  # To store the class of each shapelet
-        self._source_series = []  # To store the index of the original time series
-        self.kernels_generators_ = {}  # Reference time series
+        self._kernels = None
+        self._kernel_orig = None
+        self._start_points = []
+        self._classes = []
+        self._source_series = []
+        self.kernels_generators_ = {}
         self.n_jobs = n_jobs
+
+        # Store the seed parameter (required for sklearn compatibility)
         self.seed = seed
 
+        # Handle deprecated seed parameter
+        if seed is not None:
+            import warnings
+
+            warnings.warn(
+                "The 'seed' parameter is deprecated and will be removed in v1.2. "
+                "Use 'random_state' instead.",
+                FutureWarning,
+                stacklevel=2,
+            )
+            if random_state is None:
+                random_state = seed
+            else:
+                raise ValueError(
+                    "Cannot specify both 'seed' and 'random_state'. "
+                    "Use 'random_state' only."
+                )
+
+        self.random_state = random_state
+
     def _fit(self, X: np.ndarray, y: Union[np.ndarray, list]) -> "SAST":
         """Select reference time series and generate subsequences from them.
 
@@ -135,9 +168,9 @@ def _fit(self, X: np.ndarray, y: Union[np.ndarray, list]) -> "SAST":
         )
 
         self._random_state = (
-            np.random.RandomState(self.seed)
-            if not isinstance(self.seed, np.random.RandomState)
-            else self.seed
+            np.random.RandomState(self.random_state)
+            if not isinstance(self.random_state, np.random.RandomState)
+            else self.random_state
         )
 
         classes = np.unique(y)