[MNT] stop segmenters changing state in predict

aeon/segmentation/_fluss.py

@@ -44,7 +44,7 @@ class FLUSSSegmenter(BaseSegmenter):
     """
 
     _tags = {
-        "fit_is_empty": True,
+        "fit_is_empty": False,
         "python_dependencies": "stumpy",
     }
 
@@ -54,6 +54,15 @@ def __init__(self, period_length=10, n_regimes=2, exclusion_factor=5):
         self.exclusion_factor = exclusion_factor
         super().__init__(n_segments=n_regimes, axis=1)
 
+    def _fit(self, X, y=None):
+        if self.n_regimes < 2:
+            raise ValueError(
+                "The number of regimes must be set to an integer greater than 1"
+            )
+
+        X = X.squeeze()
+        self.found_cps, self.profiles, self.scores = self._run_fluss(X)
+
     def _predict(self, X: np.ndarray):
         """Create annotations on test/deployment data.
 
@@ -67,13 +76,6 @@ def _predict(self, X: np.ndarray):
         list
             List of change points found in X.
         """
-        if self.n_regimes < 2:
-            raise ValueError(
-                "The number of regimes must be set to an integer greater than 1"
-            )
-
-        X = X.squeeze()
-        self.found_cps, self.profiles, self.scores = self._run_fluss(X)
         return self.found_cps
 
     def predict_scores(self, X):

aeon/segmentation/_ggs.py

@@ -502,12 +502,10 @@ def _predict(self, X):
             dimension of X. The numerical values represent distinct segments
             labels for each of the data points.
         """
-        self.change_points_ = self.ggs.find_change_points(X)
+        change_points_ = self.ggs.find_change_points(X)
 
         labels = np.zeros(X.shape[0], dtype=np.int32)
-        for i, (start, stop) in enumerate(
-            zip(self.change_points_[:-1], self.change_points_[1:])
-        ):
+        for i, (start, stop) in enumerate(zip(change_points_[:-1], change_points_[1:])):
             labels[start:stop] = i
         return labels
 

aeon/segmentation/_igts.py

@@ -106,7 +106,7 @@ class _IGTS:
     """
     Information Gain based Temporal Segmentation (GTS).
 
-    GTS is a n unsupervised method for segmenting multivariate time series
+    GTS is an unsupervised method for segmenting multivariate time series
     into non-overlapping segments by locating change points that for which
     the information gain is maximized.
 
@@ -127,18 +127,13 @@ class _IGTS:
 
     Parameters
     ----------
-    k_max: int, default=10
+    k_max : int, default=10
         Maximum number of change points to find. The number of segments is thus k+1.
-    step: : int, default=5
+    step : int, default=5
         Step size, or stride for selecting candidate locations of change points.
         Fox example a `step=5` would produce candidates [0, 5, 10, ...]. Has the same
         meaning as `step` in `range` function.
 
-    Attributes
-    ----------
-    intermediate_results_: list of `ChangePointResult`
-        Intermediate segmentation results for each k value, where k=1, 2, ..., k_max
-
     Notes
     -----
     Based on the work from [1]_.
@@ -366,9 +361,9 @@ def _predict(self, X, y=None) -> np.ndarray:
             The numerical values represent distinct segment labels for each of the
             data points.
         """
-        self.change_points_ = self._igts.find_change_points(X)
-        self.intermediate_results_ = self._igts.intermediate_results_
-        return self.to_clusters(self.change_points_[1:-1], X.shape[0])
+        change_points_ = self._igts.find_change_points(X)
+        # self.intermediate_results_ = self._igts.intermediate_results_
+        return self.to_clusters(change_points_[1:-1], X.shape[0])
 
     def __repr__(self) -> str:
         """Return a string representation of the estimator."""

aeon/segmentation/tests/test_igts.py

@@ -51,5 +51,3 @@ def test_InformationGainSegmenter(multivariate_mean_shift):
             [0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3], dtype=int
         ),
     )
-    assert igts.change_points_ == [0, 5, 10, 15, 20]
-    assert len(igts.intermediate_results_) == 3

aeon/testing/testing_config.py

@@ -51,12 +51,9 @@
     "SAST": ["check_fit_deterministic"],
     "RSAST": ["check_fit_deterministic"],
     # missed in legacy testing, changes state in predict/transform
-    "FLUSSSegmenter": ["check_non_state_changing_method"],
-    "InformationGainSegmenter": ["check_non_state_changing_method"],
-    "GreedyGaussianSegmenter": ["check_non_state_changing_method"],
-    "ClaSPSegmenter": ["check_non_state_changing_method"],
     "HMMSegmenter": ["check_non_state_changing_method"],
     "RSTSF": ["check_non_state_changing_method"],
+    "ClaSPSegmenter": ["check_non_state_changing_method"],
     # Keeps length during predict to avoid recomputing means and std of data in fit
     # if the next predict calls uses the same query length parameter.
     "QuerySearch": ["check_non_state_changing_method"],