sliceline 0.2.14__tar.gz → 0.2.15__tar.gz

{sliceline-0.2.14 → sliceline-0.2.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sliceline
-Version: 0.2.14
+Version: 0.2.15
 Summary: ✂️ Fast slice finding for Machine Learning model debugging.
 Home-page: https://github.com/DataDome/sliceline
 License: BSD-3-Clause
@@ -61,7 +61,7 @@ for a more thorough tutorial:
 🛠 Installation
 ---------------
 
-Sliceline is intended to work with **Python 3.7 or above**. Installation
+Sliceline is intended to work with **Python 3.9 or above**. Installation
 can be done with ``pip``:
 
 .. code:: sh

{sliceline-0.2.14 → sliceline-0.2.15}/README.rst

@@ -41,7 +41,7 @@ for a more thorough tutorial:
 🛠 Installation
 ---------------
 
-Sliceline is intended to work with **Python 3.7 or above**. Installation
+Sliceline is intended to work with **Python 3.9 or above**. Installation
 can be done with ``pip``:
 
 .. code:: sh

{sliceline-0.2.14 → sliceline-0.2.15}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "sliceline"
-version = "0.2.14" # This version is a generic placeholder. It should not be changed.
+version = "0.2.15" # This version is a generic placeholder. It should not be changed.
 description = "✂️ Fast slice finding for Machine Learning model debugging."
 authors = ["Antoine de Daran"]
 readme = "README.rst"
@@ -22,7 +22,7 @@ scipy = "^1.12"
 black = "^24"
 flake8 = "^6"
 jupyter = "^1.0.0"
-matplotlib = "^3.8"
+matplotlib = "^3.9"
 nbconvert = "^6.5.0"
 optbinning = "^0.15.0"
 pandas = "^1.5"

{sliceline-0.2.14 → sliceline-0.2.15}/sliceline/slicefinder.py

@@ -378,7 +378,11 @@ class Slicefinder(BaseEstimator, TransformerMixin):
         slice_candidates = x_encoded @ slices.T == level
         slice_sizes = slice_candidates.sum(axis=0).A[0]
         slice_errors = errors @ slice_candidates
-        max_slice_errors = slice_candidates.T.multiply(errors).max(axis=1).A
+        # Here we can't use the .A shorthand because it is not
+        # implemented in all scipy versions for coo_matrix objects
+        max_slice_errors = (
+            slice_candidates.T.multiply(errors).max(axis=1).toarray()
+        )
 
         # score of relative error and relative size
         slice_scores = self._score(
@@ -397,7 +401,11 @@ class Slicefinder(BaseEstimator, TransformerMixin):
         """Initialise 1-slices, i.e. slices with one predicate."""
         slice_sizes = x_encoded.sum(axis=0).A[0]
         slice_errors = errors @ x_encoded
-        max_slice_errors = x_encoded.T.multiply(errors).max(axis=1).A[:, 0]
+        # Here we can't use the .A shorthand because it is not
+        # implemented in all scipy versions for coo_matrix objects
+        max_slice_errors = (
+            x_encoded.T.multiply(errors).max(axis=1).toarray()[:, 0]
+        )
 
         # working set of active slices (#attr x #slices) and top-k
         valid_slices_mask = (slice_sizes >= self.min_sup) & (slice_errors > 0)
@@ -440,7 +448,9 @@ class Slicefinder(BaseEstimator, TransformerMixin):
     ) -> np.ndarray:
         """Join compatible slices according to `level`."""
         slices_int = slices.astype(int)
-        join = (slices_int @ slices_int.T).A == level - 2
+        # Here we can't use the .A shorthand because it is not
+        # implemented in all scipy versions for coo_matrix objects
+        join = (slices_int @ slices_int.T).toarray() == level - 2
         return np.triu(join, 1) * join
 
     @staticmethod
@@ -503,7 +513,10 @@ class Slicefinder(BaseEstimator, TransformerMixin):
             sub_pair_candidates = pair_candidates[:, start:end]
             # sub_p should not contain multiple True on the same line
             i = sub_pair_candidates.argmax(axis=1).T + np.any(
-                sub_pair_candidates.A, axis=1
+                # Here we can't use the .A shorthand because it is not
+                # implemented in all scipy versions for coo_matrix objects
+                sub_pair_candidates.toarray(),
+                axis=1,
             )
             ids = ids + i.A * np.prod(dom[(j + 1) : dom.shape[0]])
         return ids