[ENH] Include unconstrained DTW and DDTW in ElasticEnsemble parameters

aeon/classification/distance_based/_elastic_ensemble.py

@@ -413,7 +413,9 @@ def get_inclusive(min_val: float, max_val: float, num_vals: float):
             return np.arange(min_val, max_val + inc / 2, inc)
 
         if distance_measure == "dtw" or distance_measure == "ddtw":
-            return {"distance_params": [{"window": x / 100} for x in range(0, 100)]}
+            constrained_params = [{"window": x / 100} for x in range(1, 100)]
+            unconstrained_param = {"window": 1.0}
+            return {"distance_params": [unconstrained_param] + constrained_params}
         elif distance_measure == "wdtw" or distance_measure == "wddtw":
             return {"distance_params": [{"g": x / 100} for x in range(0, 100)]}
         elif distance_measure == "lcss":

aeon/classification/distance_based/tests/test_elastic_ensemble.py

@@ -90,3 +90,18 @@ def test_all_distance_measures():
     ee.fit(X, y)
     distances = list(ee.get_metric_params())
     assert len(distances) == 9
+
+
+@pytest.mark.parametrize("data", [DATA])
+def test_dtw_ddtw_full_window_inclusion(data):
+    """Test that window=1.0 (full window) is included in DTW/DDTW parameters."""
+    dtw_paras = ElasticEnsemble._get_100_param_options("dtw", data)
+    dtw_param_values = dtw_paras["distance_params"]
+    dtw_windows = {p.get("window") for p in dtw_param_values}
+    assert 1.0 in dtw_windows
+    assert len(dtw_param_values) == 100
+    ddtw_paras = ElasticEnsemble._get_100_param_options("ddtw", data)
+    ddtw_param_values = ddtw_paras["distance_params"]
+    ddtw_windows = {p.get("window") for p in ddtw_param_values}
+    assert 1.0 in ddtw_windows
+    assert len(ddtw_param_values) == 100

docs/changelogs/v1.3.md

@@ -39,6 +39,7 @@ September 2025
 - [ENH] KNN n_jobs and updated kneighbours method ({pr}`2578`) {user}`chrisholder`
 - [ENH] Refactor signature code ({pr}`2943`) {user}`TonyBagnall`
 - [ENH] Change seed to random_state ({pr}`3031`) {user}`TonyBagnall`
+- [ENH] Include unconstrained DTW and DDTW in Elastic Ensemble ({pr}`3112`) {user}`Nithurshen`
 
 ## Clustering