[ENH] Refactor example dataset for unequal length univariate (#2859)

* deprecate PLAID

* add pickup

* deprecated

Author: TonyBagnall

aeon/datasets/__init__.py

@@ -25,6 +25,7 @@
     "load_osuleaf",
     "load_italy_power_demand",
     "load_japanese_vowels",
+    "load_pickup_gesture_wiimoteZ",
     "load_plaid",
     "load_longley",
     "load_lynx",
@@ -74,6 +75,7 @@
     load_lynx,
     load_osuleaf,
     load_PBS_dataset,
+    load_pickup_gesture_wiimoteZ,
     load_plaid,
     load_shampoo_sales,
     load_solar,

aeon/datasets/_single_problem_loaders.py

@@ -28,6 +28,7 @@
 
 import numpy as np
 import pandas as pd
+from deprecated.sphinx import deprecated
 
 from aeon.datasets import load_from_tsf_file
 from aeon.datasets._data_loaders import _load_saved_dataset, _load_tsc_dataset
@@ -344,6 +345,48 @@ def load_basic_motions(split=None, return_type="numpy3d"):
     )
 
 
+def load_pickup_gesture_wiimoteZ(split=None):
+    """Load the PickupGestureWiimoteZ univariate time series classification problem.
+
+    Example of a univariate problem with unequal length time series.
+
+    Parameters
+    ----------
+    split: None or one of "TRAIN", "TEST", default=None
+        Whether to load the train or test instances of the problem. By default it
+        loads both train and test instances into a single array.
+
+    Returns
+    -------
+    X: list of 2D np.ndarray, one for each series.
+    y: 1D numpy array of length len(X). The class labels for each time series
+    instance in X.
+
+    Notes
+    -----
+    Dimensionality:     univariate
+    Series length:      variable
+    Train cases:        50
+    Test cases:         50
+    Number of classes:  2
+    https://timeseriesclassification.com/description.php?Dataset=PickupGestureWiimoteZ
+
+    Examples
+    --------
+    >>> from aeon.datasets import load_pickup_gesture_wiimoteZ
+    >>> X, y = load_pickup_gesture_wiimoteZ()
+    """
+    return _load_tsc_dataset("PickupGestureWiimoteZ", split, return_type="np-list")
+
+
+# TODO: remove in v1.3.0
+@deprecated(
+    version="1.2.0",
+    reason="load_plaid and the PLAID dataset will be removed in version 1.3. It has "
+    "been replaced by a smaller unequal length univariate problem "
+    "PickupGestureWiimoteZ.",
+    category=FutureWarning,
+)
 def load_plaid(split=None):
     """Load the PLAID univariate time series classification problem.
 

aeon/datasets/data/PickupGestureWiimoteZ/PickupGestureWiimoteZ_TEST.ts

@@ -318,7 +318,8 @@ def test_load_from_ts_file():
     Test
     1. Univariate equal length (UnitTest) returns 3D numpy X, 1D numpy y
     2. Multivariate equal length (BasicMotions) returns 3D numpy X, 1D numpy y
-    3. Univariate and multivariate unequal length (PLAID) return X as list of numpy
+    3. unequal length Univariate (PickupGestureWiimoteZ) and multivariate (
+    JapaneseVowels) return X as list of numpy
     """
     # Test 1.1: load univariate equal length (UnitTest), should return 2D array and 1D
     # array, test first and last data
@@ -338,7 +339,8 @@ def test_load_from_ts_file():
     assert X.shape == (20, 24)
     assert X[0][0] == 573.0
 
-    # Test 2: load multivare equal length (BasicMotions), should return 3D array and 1D
+    # Test 2: load multivariate equal length (BasicMotions), should return 3D array
+    # and 1D
     # array, test first and last data.
     data_path = os.path.join(
         os.path.dirname(aeon.__file__),
@@ -354,16 +356,16 @@ def test_load_from_ts_file():
     assert X.shape == (40, 6, 100) and y.shape == (40,)
     assert X[1][2][3] == -1.898794
 
-    # Test 3.1: load univariate unequal length (PLAID), should return a one column
-    # dataframe,
+    # Test 3.1: load univariate unequal length (PickupGestureWiimoteZ), should return
+    # a list up numpy arrays
     data_path = os.path.join(
         os.path.dirname(aeon.__file__),
-        "datasets/data/PLAID/PLAID_TRAIN.ts",
+        "datasets/data/PickupGestureWiimoteZ/PickupGestureWiimoteZ_TRAIN.ts",
     )
 
     X, y = load_from_ts_file(full_file_path_and_name=data_path, return_meta_data=False)
     assert isinstance(X, list) and isinstance(y, np.ndarray)
-    assert len(X) == 537 and y.shape == (537,)
+    assert len(X) == 50 and y.shape == (50,)
     # Test 3.2: load multivariate unequal length (JapaneseVowels), should return a X
     # columns dataframe,
     data_path = os.path.join(

aeon/datasets/data/PickupGestureWiimoteZ/PickupGestureWiimoteZ_TRAIN.ts

@@ -20,7 +20,7 @@
     load_lynx,
     load_osuleaf,
     load_PBS_dataset,
-    load_plaid,
+    load_pickup_gesture_wiimoteZ,
     load_shampoo_sales,
     load_solar,
     load_unit_test,
@@ -39,7 +39,7 @@
     load_basic_motions,
 ]
 UNEQUAL_LENGTH_PROBLEMS = [
-    load_plaid,
+    load_pickup_gesture_wiimoteZ,
     load_japanese_vowels,
 ]
 

aeon/datasets/data/PickupGestureWiimoteZ/PickupGestureWiimoteZ_eq_TEST.ts

@@ -6,14 +6,20 @@
 from aeon.datasets import (
     load_basic_motions,
     load_japanese_vowels,
-    load_plaid,
+    load_pickup_gesture_wiimoteZ,
     load_unit_test,
 )
 from aeon.transformations.collection import Padder
 
 
 @pytest.mark.parametrize(
-    "loader", [load_japanese_vowels, load_plaid, load_unit_test, load_basic_motions]
+    "loader",
+    [
+        load_japanese_vowels,
+        load_pickup_gesture_wiimoteZ,
+        load_unit_test,
+        load_basic_motions,
+    ],
 )
 def test_padding(loader):
     """Test padding to on provided datasets."""

aeon/datasets/data/PickupGestureWiimoteZ/PickupGestureWiimoteZ_eq_TRAIN.ts

@@ -6,14 +6,20 @@
 from aeon.datasets import (
     load_basic_motions,
     load_japanese_vowels,
-    load_plaid,
+    load_pickup_gesture_wiimoteZ,
     load_unit_test,
 )
 from aeon.transformations.collection import Truncator
 
 
 @pytest.mark.parametrize(
-    "loader", [load_japanese_vowels, load_plaid, load_unit_test, load_basic_motions]
+    "loader",
+    [
+        load_japanese_vowels,
+        load_pickup_gesture_wiimoteZ,
+        load_unit_test,
+        load_basic_motions,
+    ],
 )
 def test_truncation_transformer(loader):
     """Test truncation to the fixed series length on provided datasets."""