Improved coverage. Back to 100% for FLOSS

Author: seanlaw

docs/Tutorial_Semantic_Segmentation.ipynb

@@ -166,7 +166,7 @@
     {
      "data": {
       "text/plain": [
-       "<matplotlib.patches.Rectangle at 0x1a1b22ac50>"
+       "<matplotlib.patches.Rectangle at 0x1a1af70c50>"
       ]
      },
      "execution_count": 4,
@@ -216,7 +216,7 @@
     {
      "data": {
       "text/plain": [
-       "<matplotlib.patches.FancyArrowPatch at 0x1a1bbfbe50>"
+       "<matplotlib.patches.FancyArrowPatch at 0x1a1b942f50>"
       ]
      },
      "execution_count": 5,
@@ -336,7 +336,7 @@
     {
      "data": {
       "text/plain": [
-       "<matplotlib.lines.Line2D at 0x1a1eeeaa50>"
+       "<matplotlib.lines.Line2D at 0x1a1ecba590>"
       ]
      },
      "execution_count": 8,
@@ -397,7 +397,7 @@
     {
      "data": {
       "text/plain": [
-       "[<matplotlib.lines.Line2D at 0x1a1f5a1210>]"
+       "[<matplotlib.lines.Line2D at 0x1a1f6e48d0>]"
       ]
      },
      "execution_count": 10,
@@ -491,8 +491,9 @@
     "You can now update the `stream` with a new data points `t` via the `stream.update(t)` function and this will slide your window over by one data point and it will automatically update:\n",
     "\n",
     "1. the `CAC_1D` (accessed via the `.cac_` attribute)\n",
-    "2. the matrix profile (accessed via the `.mp_` attribute)\n",
-    "3. the sliding window of data used to produce the `CAC_1D` (accessed via the `.T_` attribute - this should be the same size as the length of the `old_data\\)\n",
+    "2. the matrix profile (accessed via the `.P_` attribute)\n",
+    "3. the matrix profile indices (accessed via the `.I_` attribute)\n",
+    "4. the sliding window of data used to produce the `CAC_1D` (accessed via the `.T_` attribute - this should be the same size as the length of the `old_data\\)\n",
     "\n",
     "Let's continuosly update our `stream` with the `new_data` one value at a time and store them in a list (you'll see why in a second):"
    ]

stumpy/floss.py

@@ -341,12 +341,13 @@ class floss(object):
         A corrected arc curve (CAC) updated as a result of ingressing a single
         new data point and egressing a single old data point.
 
-    mp_ : ndarray
-        The first column consists of the matrix profile and the second column
-        consists of the matrix profile indices. This updated matrix profile and
-        matrix profile indices are a result of ingressing a single new data
-        point and egressing a single old data point. Note that this array does
-        not contain the left and right matrix profile indices.
+    P_ : ndarray
+        The matrix profile updated as a result of ingressing a single new data
+        point and egressing a single old data point.
+
+    I_ : ndarray
+        The (right) matrix profile indices updated as a result of ingressing a single
+        new data point and egressing a single old data point.
 
     T_ : ndarray
         The updated time series, `T`
@@ -519,18 +520,25 @@ def update(self, t):
         self._n_appended += 1
 
     @property
-    def mp_(self):
+    def cac_(self):
+        """
+        Get the updated corrected arc curve (CAC)
+        """
+        return self._cac.astype(np.float)
+
+    @property
+    def P_(self):
         """
         Get the updated matrix profile
         """
-        return self._mp.astype(np.float)
+        return self._mp[:, 0].astype(np.float)
 
     @property
-    def cac_(self):
+    def I_(self):
         """
-        Get the updated corrected arc curve (CAC)
+        Get the updated (right) matrix profile indices
         """
-        return self._cac.astype(np.float)
+        return self._mp[:, 3].astype(np.float)
 
     @property
     def T_(self):

tests/test_floss.py

@@ -3,6 +3,7 @@
 from stumpy import _nnmark, _iac, _cac, _rea, fluss, stump, core, floss
 import copy
 import pytest
+import naive
 
 
 def naive_nnmark(I):
@@ -141,9 +142,9 @@ def test_floss():
     n = old_data.shape[0]
     add_data = data[30:]
 
-    left_mp = naive_right_mp(old_data, m)
+    mp = naive_right_mp(old_data, m)
     right_mp = stump(old_data, m)
-    k = left_mp.shape[0]
+    k = mp.shape[0]
 
     rolling_Ts = core.rolling_window(data[1:], n)
     L = 5
@@ -153,25 +154,42 @@ def test_floss():
     last_idx = n - m + 1
     excl_zone = int(np.ceil(m / 4))
     zone_start = max(0, k - excl_zone)
-    for i, T in enumerate(rolling_Ts):
-        left_mp[:] = np.roll(left_mp, -1, axis=0)
-        left_mp[-1, 0] = np.inf
-        left_mp[-1, 3] = last_idx + i
-
-        D = naive_distance_profile(T[-m:], T, m)
+    for i, left_T in enumerate(rolling_Ts):
+        mp[:, 1] = -1
+        mp[:, 2] = -1
+        mp[:] = np.roll(mp, -1, axis=0)
+        mp[-1, 0] = np.inf
+        mp[-1, 3] = last_idx + i
+
+        D = naive_distance_profile(left_T[-m:], left_T, m)
         D[zone_start:] = np.inf
 
-        update_idx = np.argwhere(D < left_mp[:, 0]).flatten()
-        left_mp[update_idx, 0] = D[update_idx]
-        left_mp[update_idx, 3] = last_idx + i
+        update_idx = np.argwhere(D < mp[:, 0]).flatten()
+        mp[update_idx, 0] = D[update_idx]
+        mp[update_idx, 3] = last_idx + i
 
         left_cac = _cac(
-            left_mp[:, 3] - i - 1,
+            mp[:, 3] - i - 1,
             L,
             bidirectional=False,
             excl_factor=excl_factor,
             custom_iac=custom_iac,
         )
-        stream.update(T[-1])
+
+        left_mp = mp.copy()
+        left_P = left_mp[:, 0]
+        left_I = left_mp[:, 3]
+
+        stream.update(left_T[-1])
         right_cac = stream.cac_
+        right_P = stream.P_
+        right_I = stream.I_
+        right_T = stream.T_
+
+        naive.replace_inf(left_P)
+        naive.replace_inf(right_P)
+
         npt.assert_almost_equal(left_cac, right_cac)
+        npt.assert_almost_equal(left_P, right_P)
+        npt.assert_almost_equal(left_I, right_I)
+        npt.assert_almost_equal(left_T, right_T)