Fixed #966 Add Named Attributes to Matrix Profile

stumpy/aamp.py

@@ -7,6 +7,7 @@
 from numba import njit, prange
 
 from . import config, core
+from .mparray import mparray
 
 
 @njit(
@@ -378,6 +379,12 @@ def aamp(T_A, m, T_B=None, ignore_trivial=True, p=2.0, k=1):
         equivalently, out[:, -2] and out[:, -1]) correspond to the top-1 left
         matrix profile indices and the top-1 right matrix profile indices, respectively.
 
+        For convenience, the matrix profile (distances) and matrix profile indices can
+        also be accessed via their corresponding named array attributes, `.P_` and
+        `.I_`,respectively. Similarly, the corresponding left matrix profile indices
+        and right matrix profile indices may also be accessed via the `.left_I_` and
+        `.right_I_` array attributes.
+
     Notes
     -----
     `arXiv:1901.05708 \
@@ -431,4 +438,4 @@ def aamp(T_A, m, T_B=None, ignore_trivial=True, p=2.0, k=1):
 
     core._check_P(out[:, 0])
 
-    return out
+    return mparray(out, m=m, k=k)

stumpy/aamped.py

@@ -7,6 +7,7 @@
 
 from . import config, core
 from .aamp import _aamp
+from .mparray import mparray
 
 
 def _dask_aamped(
@@ -212,6 +213,12 @@ def aamped(client, T_A, m, T_B=None, ignore_trivial=True, p=2.0, k=1):
         equivalently, out[:, -2] and out[:, -1]) correspond to the top-1 left
         matrix profile indices and the top-1 right matrix profile indices, respectively.
 
+        For convenience, the matrix profile (distances) and matrix profile indices can
+        also be accessed via their corresponding named array attributes, `.P_` and
+        `.I_`,respectively. Similarly, the corresponding left matrix profile indices
+        and right matrix profile indices may also be accessed via the `.left_I_` and
+        `.right_I_` array attributes.
+
     Notes
     -----
     `arXiv:1901.05708 \
@@ -264,4 +271,4 @@ def aamped(client, T_A, m, T_B=None, ignore_trivial=True, p=2.0, k=1):
 
     core._check_P(out[:, 0])
 
-    return out
+    return mparray(out, m=m, k=k)

stumpy/chains.py

@@ -52,6 +52,11 @@ def atsc(IL, IR, j):
     >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
     >>> stumpy.atsc(mp[:, 2], mp[:, 3], 1)
     array([1, 3])
+
+    # Alternative example using named attributes
+    >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
+    >>> stumpy.atsc(mp.left_I_, mp.right_I_, 1)
+    array([1, 3])
     """
     C = deque([j])
     for i in range(IL.size):
@@ -118,6 +123,11 @@ def allc(IL, IR):
     >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
     >>> stumpy.allc(mp[:, 2], mp[:, 3])
     ([array([1, 3]), array([2]), array([0, 4])], array([0, 4]))
+
+    # Alternative example using named attributes
+    >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
+    >>> stumpy.allc(mp.left_I_, mp.right_I_)
+    ([array([1, 3]), array([2]), array([0, 4])], array([0, 4]))
     """
     L = np.ones(IL.size, dtype=np.int64)
     S = set()  # type: ignore

stumpy/floss.py

@@ -290,6 +290,11 @@ def fluss(I, L, n_regimes, excl_factor=5, custom_iac=None):
     >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
     >>> stumpy.fluss(mp[:, 0], 3, 2)
     (array([1., 1., 1., 1., 1.]), array([0]))
+
+    # Alternative example using named attributes
+    >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
+    >>> stumpy.fluss(mp.P_, 3, 2)
+    (array([1., 1., 1., 1., 1.]), array([0]))
     """
     cac = _cac(I, L, bidirectional=True, excl_factor=excl_factor, custom_iac=custom_iac)
     regime_locs = _rea(cac, n_regimes, L, excl_factor=excl_factor)

stumpy/gpu_aamp.py

@@ -9,6 +9,7 @@
 from numba import cuda
 
 from . import config, core
+from .mparray import mparray
 
 
 @cuda.jit(
@@ -496,6 +497,12 @@ def gpu_aamp(T_A, m, T_B=None, ignore_trivial=True, device_id=0, p=2.0, k=1):
         equivalently, out[:, -2] and out[:, -1]) correspond to the top-1 left
         matrix profile indices and the top-1 right matrix profile indices, respectively.
 
+        For convenience, the matrix profile (distances) and matrix profile indices can
+        also be accessed via their corresponding named array attributes, `.P_` and
+        `.I_`,respectively. Similarly, the corresponding left matrix profile indices
+        and right matrix profile indices may also be accessed via the `.left_I_` and
+        `.right_I_` array attributes.
+
     Notes
     -----
     `arXiv:1901.05708 \
@@ -706,4 +713,4 @@ def gpu_aamp(T_A, m, T_B=None, ignore_trivial=True, device_id=0, p=2.0, k=1):
 
     core._check_P(out[:, 0])
 
-    return out
+    return mparray(out, m=m, k=k)

stumpy/gpu_stump.py

@@ -10,6 +10,7 @@
 
 from . import config, core
 from .gpu_aamp import gpu_aamp
+from .mparray import mparray
 
 
 @cuda.jit(
@@ -573,6 +574,12 @@ def gpu_stump(
         equivalently, out[:, -2] and out[:, -1]) correspond to the top-1 left
         matrix profile indices and the top-1 right matrix profile indices, respectively.
 
+        For convenience, the matrix profile (distances) and matrix profile indices can
+        also be accessed via their corresponding named array attributes, `.P_` and
+        `.I_`,respectively. Similarly, the corresponding left matrix profile indices
+        and right matrix profile indices may also be accessed via the `.left_I_` and
+        `.right_I_` array attributes. See examples below.
+
     See Also
     --------
     stumpy.stump : Compute the z-normalized matrix profile
@@ -615,15 +622,20 @@ def gpu_stump(
     >>> from numba import cuda
     >>> if __name__ == "__main__":
     ...     all_gpu_devices = [device.id for device in cuda.list_devices()]
-    ...     stumpy.gpu_stump(
+    ...     mp = stumpy.gpu_stump(
     ...         np.array([584., -11., 23., 79., 1001., 0., -19.]),
     ...         m=3,
     ...         device_id=all_gpu_devices)
-    array([[0.11633857113691416, 4, -1, 4],
-           [2.694073918063438, 3, -1, 3],
-           [3.0000926340485923, 0, 0, 4],
-           [2.694073918063438, 1, 1, -1],
-           [0.11633857113691416, 0, 0, -1]], dtype=object)
+    >>>     mp
+    mparray([[0.11633857113691416, 4, -1, 4],
+             [2.694073918063438, 3, -1, 3],
+             [3.0000926340485923, 0, 0, 4],
+             [2.694073918063438, 1, 1, -1],
+             [0.11633857113691416, 0, 0, -1]], dtype=object)
+    >>>     mp.P_
+    mparray([0.11633857, 2.69407392, 3.00009263, 2.69407392, 0.11633857])
+    >>>     mp.I_
+    mparray([4, 3, 0, 1, 0])
     """
     if T_B is None:  # Self join!
         T_B = T_A
@@ -838,4 +850,4 @@ def gpu_stump(
 
     core._check_P(out[:, 0])
 
-    return out
+    return mparray(out, m=m, k=k)

stumpy/motifs.py

@@ -298,6 +298,14 @@ def motifs(
     ...     mp[:, 0],
     ...     max_distance=2.0)
     (array([[0.        , 0.11633857]]), array([[0, 4]]))
+
+    # Alternative example using named attributes
+    >>> mp = stumpy.stump(np.array([584., -11., 23., 79., 1001., 0., -19.]), m=3)
+    >>> stumpy.motifs(
+    ...     np.array([584., -11., 23., 79., 1001., 0., -19.]),
+    ...     mp.P_,
+    ...     max_distance=2.0)
+    (array([[0.        , 0.11633857]]), array([[0, 4]]))
     """
     T = core._preprocess(T)
 

stumpy/mparray.py

@@ -0,0 +1,184 @@
+import numpy as np
+
+
+class mparray(np.ndarray):
+    """
+    A matrix profile convenience class that subclasses the numpy ndarray
+
+    Parameters
+    ----------
+    cls : class
+        The base class
+
+    input_array : ndarray
+        The input `numpy` array to be sublcassed
+
+    m : int
+        Window size
+
+    k : int, default 1
+        The number of top `k` smallest distances used to construct the
+        matrix profile.
+
+    Attributes
+    ----------
+    P_ : numpy.ndarray
+        The (top-k) matrix profile for `T`. When `k=1` (default), the first
+        (and only) column in this 2D array, which consists of the matrix profile,
+        is returned. When `k > 1`, the output has exactly `k` columns consisting of
+        the top-k matrix profile.
+
+    I_ : numpy.ndarray
+        The(top-k) matrix profile indices for `T`. When `k=1` (default), the first
+        (and only) column in this 2D array, which consists of the matrix profile,
+        indices is returned. When `k > 1`, the output has exactly `k` columns
+        consisting of the top-k matrix profile indices.
+
+    left_P_ : numpy.ndarray
+        The left (top-1) matrix profile for `T`
+
+    left_I_ : numpy.ndarray
+        The updated left (top-1) matrix profile indices for `T`
+    """
+
+    def __new__(cls, input_array, m, k):
+        """
+        Create the ndarray instance of our type, given the usual
+        ndarray input arguments.  This will call the standard
+        ndarray constructor, but return an object of our type.
+        It also triggers a call mparray.__array_finalize__
+
+        Parameters
+        ----------
+        cls : class
+            The base class
+
+        input_array : ndarray
+            The input `numpy` array to be sublcassed
+
+        m : int
+            Window size
+
+        k : int
+            The number of top `k` smallest distances used to construct the
+            matrix profile.
+        """
+        obj = np.asarray(input_array).view(cls)
+        obj._m = m
+        obj._k = k
+        # All new attributes will also needed to be added to the `__array_finalize__`
+        # function below so that "new-from-template" objects (e.g., an array slice)
+        # will also contain the same new attributes
+        return obj
+
+    def __array_finalize__(self, obj):
+        """
+        Finalize the array
+
+        Parameters
+        ----------
+        obj : object
+            This is the class object
+        """
+        if obj is None:  # pragma: no cover
+            return
+        # The lines below ensure that child objects that are created from a slice
+        # of an `mparray` will also inherit the attributes from the parent `mparray`
+        self._m = getattr(obj, "_m", None)
+        self._k = getattr(obj, "_k", None)
+
+    def _P(self):
+        """
+        Matrix profile values
+
+        Parameters
+        ----------
+        None
+        """
+        if self._k == 1:
+            return self[:, : self._k].flatten().astype(np.float64)
+        else:
+            return self[:, : self._k].astype(np.float64)
+
+    def _I(self):
+        """
+        Nearest neighbor indices
+
+        Parameters
+        ----------
+        None
+        """
+        if self._k == 1:
+            return self[:, self._k : 2 * self._k].flatten().astype(np.int64)
+        else:
+            return self[:, self._k : 2 * self._k].astype(np.int64)
+
+    def _left_I(self):
+        """
+        Left nearest neighbor indices
+
+        Parameters
+        ----------
+        None
+        """
+        if self._k == 1:
+            return self[:, 2 * self._k].flatten().astype(np.int64)
+        else:
+            return self[:, 2 * self._k].astype(np.int64)
+
+    def _right_I(self):
+        """
+        Right nearest neighbor indices
+
+        Parameters
+        ----------
+        None
+        """
+        if self._k == 1:
+            return self[:, 2 * self._k + 1].flatten().astype(np.int64)
+        else:
+            return self[:, 2 * self._k + 1].astype(np.int64)
+
+    @property
+    def P_(self):
+        """
+        Matrix profile values
+
+        Parameters
+        ----------
+        None
+        """
+        return self._P()
+
+    @property
+    def I_(self):
+        """
+        Nearest neighbor indices
+
+        Parameters
+        ----------
+        None
+        """
+        return self._I()
+
+    @property
+    def left_I_(self):
+        """
+        Left nearest neighbor indices
+
+        Parameters
+        ----------
+        None
+        """
+        return self._left_I()
+
+    @property
+    def right_I_(self):
+        """
+        Right nearest neighbor indices
+
+        Parameters
+        ----------
+        None
+        """
+        return self._right_I()