[ENH] Collection conversion cleanup and df-list fix (#2654)

* collection conversion cleanup

* notebook

* fixes

---------

Co-authored-by: Tony Bagnall <[email protected]>

Author: MatthewMiddlehurst

aeon/classification/early_classification/tests/test_probability_threshold.py

@@ -32,7 +32,7 @@ def test_early_prob_threshold_near_classification_points():
         X = X_test[:, :, :i]
 
         if i == 20:
-            with pytest.raises(ValueError):
+            with pytest.raises(IndexError):
                 pt.update_predict_proba(X)
         else:
             _, decisions = pt.update_predict_proba(X)

aeon/classification/early_classification/tests/test_teaser.py

@@ -80,7 +80,7 @@ def test_teaser_near_classification_points():
         X = X_test[:, :, :i]
 
         if i == 20:
-            with pytest.raises(ValueError):
+            with pytest.raises(IndexError):
                 teaser.update_predict_proba(X)
         else:
             _, decisions = teaser.update_predict(X)

aeon/testing/data_generation/_collection.py

@@ -408,15 +408,11 @@ def make_example_dataframe_list(
     ...     random_state=0,
     ... )
     >>> print(data)
-    [          0         1
-    0  0.000000  1.688531
-    1  1.715891  1.694503
-    2  1.247127  0.768763
-    3  0.595069  0.113426,           0         1
-    0  2.000000  3.166900
-    1  2.115580  2.272178
-    2  3.702387  0.284144
-    3  0.348517  0.080874]
+    [          0         1         2         3
+    0  0.000000  1.688531  1.715891  1.694503
+    1  1.247127  0.768763  0.595069  0.113426,           0         1         2         3
+    0  2.000000  3.166900  2.115580  2.272178
+    1  3.702387  0.284144  0.348517  0.080874]
     >>> print(labels)
     [0 1]
     >>> get_type(data)
@@ -428,14 +424,14 @@ def make_example_dataframe_list(
 
     for i in range(n_cases):
         n_timepoints = rng.randint(min_n_timepoints, max_n_timepoints + 1)
-        x = n_labels * rng.uniform(size=(n_timepoints, n_channels))
+        x = n_labels * rng.uniform(size=(n_channels, n_timepoints))
         label = x[0, 0].astype(int)
         if i < n_labels and n_cases > i:
             x[0, 0] = i
             label = i
         x = x * (label + 1)
 
-        X.append(pd.DataFrame(x, index=range(n_timepoints), columns=range(n_channels)))
+        X.append(pd.DataFrame(x, index=range(n_channels), columns=range(n_timepoints)))
         y[i] = label
 
     if regression_target:
@@ -574,16 +570,16 @@ def make_example_multi_index_dataframe(
     ...     random_state=0,
     ... )
     >>> print(data)  # doctest: +NORMALIZE_WHITESPACE
-                    channel_0  channel_1
+    channel                0         1
     case timepoint
-    0    0           0.000000   1.247127
-         1           1.688531   0.768763
-         2           1.715891   0.595069
-         3           1.694503   0.113426
-    1    0           2.000000   3.702387
-         1           3.166900   0.284144
-         2           2.115580   0.348517
-         3           2.272178   0.080874
+    0    0          0.000000  1.247127
+         1          1.688531  0.768763
+         2          1.715891  0.595069
+         3          1.694503  0.113426
+    1    0          2.000000  3.702387
+         1          3.166900  0.284144
+         2          2.115580  0.348517
+         3          2.272178  0.080874
     >>> print(labels)
     [0 1]
     >>> get_type(data)
@@ -616,8 +612,7 @@ def make_example_multi_index_dataframe(
         y[i] = label
 
     X = X.reset_index(drop=True)
-    X = X.set_index(["case", "timepoint"]).pivot(columns="channel")
-    X.columns = [f"channel_{i}" for i in range(n_channels)]
+    X = X.pivot(index=["case", "timepoint"], columns=["channel"], values="value")
 
     if regression_target:
         y = y.astype(np.float32)

aeon/testing/data_generation/tests/test_collection.py

@@ -178,13 +178,13 @@ def test_make_example_dataframe_list(
     assert all(isinstance(x, pd.DataFrame) for x in X)
     assert isinstance(y, np.ndarray)
     assert len(X) == n_cases
-    assert all([x.shape[1] == n_channels for x in X])
+    assert all([x.shape[0] == n_channels for x in X])
     if min_n_timepoints == max_n_timepoints:
-        assert all([x.shape[0] == min_n_timepoints for x in X])
+        assert all([x.shape[1] == min_n_timepoints for x in X])
     else:
         assert all(
             [
-                x.shape[0] >= min_n_timepoints and x.shape[0] <= max_n_timepoints
+                x.shape[1] >= min_n_timepoints and x.shape[1] <= max_n_timepoints
                 for x in X
             ]
         )

aeon/testing/tests/test_testing_data.py

@@ -122,10 +122,9 @@ def test_equal_length_univariate_collection():
         assert not is_collection(
             EQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
-        assert is_univariate(
-            EQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
-            is_collection=False,
-        )
+        # assert is_univariate(
+        #     EQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
+        # )
         assert is_equal_length(
             EQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
@@ -199,10 +198,9 @@ def test_unequal_length_univariate_collection():
         assert not is_collection(
             UNEQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
-        assert is_univariate(
-            UNEQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
-            is_collection=False,
-        )
+        # assert is_univariate(
+        #     UNEQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
+        # )
         assert is_equal_length(
             UNEQUAL_LENGTH_UNIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
@@ -276,10 +274,9 @@ def test_equal_length_multivariate_collection():
         assert not is_collection(
             EQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
-        assert not is_univariate(
-            EQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
-            is_collection=False,
-        )
+        # assert not is_univariate(
+        #     EQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
+        # )
         assert is_equal_length(
             EQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
@@ -365,10 +362,9 @@ def test_unequal_length_multivariate_collection():
         assert not is_collection(
             UNEQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )
-        assert not is_univariate(
-            UNEQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
-            is_collection=False,
-        )
+        # assert not is_univariate(
+        #     UNEQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0],
+        # )
         assert is_equal_length(
             UNEQUAL_LENGTH_MULTIVARIATE_SIMILARITY_SEARCH[key]["test"][0]
         )

aeon/testing/utils/deep_equals.py

@@ -84,6 +84,8 @@ def _deep_equals(x, y, depth, ignore_index):
 def _series_equals(x, y, depth, ignore_index):
     if x.dtype != y.dtype:
         return False, f"x.dtype ({x.dtype}) != y.dtype ({y.dtype}), depth={depth}"
+    if x.shape != y.shape:
+        return False, f"x.shape ({x.shape}) != y.shape ({y.shape}), depth={depth}"
 
     # if columns are object, recurse over entries and index
     if x.dtype == "object":
@@ -108,7 +110,12 @@ def _series_equals(x, y, depth, ignore_index):
 
 def _dataframe_equals(x, y, depth, ignore_index):
     if not x.columns.equals(y.columns):
-        return False, f"x.columns ({x.columns}) != y.columns ({y.columns})"
+        return (
+            False,
+            f"x.columns ({x.columns}) != y.columns ({y.columns}), depth={depth}",
+        )
+    if x.shape != y.shape:
+        return False, f"x.shape ({x.shape}) != y.shape ({y.shape}), depth={depth}"
 
     # if columns are equal and at least one is object, recurse over Series
     if sum(x.dtypes == "object") > 0:
@@ -130,7 +137,9 @@ def _dataframe_equals(x, y, depth, ignore_index):
 
 def _numpy_equals(x, y, depth, ignore_index):
     if x.dtype != y.dtype:
-        return False, f"x.dtype ({x.dtype}) != y.dtype ({y.dtype})"
+        return False, f"x.dtype ({x.dtype}) != y.dtype ({y.dtype}), depth={depth}"
+    if x.shape != y.shape:
+        return False, f"x.shape ({x.shape}) != y.shape ({y.shape}), depth={depth}"
 
     if x.dtype == "object":
         for i in range(len(x)):