action-rules 2.0.0__tar.gz → 2.0.1__tar.gz

{action_rules-2.0.0 → action_rules-2.0.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: action-rules
-Version: 2.0.0
+Version: 2.0.1
 Summary: The package for action rules mining using Action-Apriori (Apriori Modified for Action Rules Mining)..
 License: MIT
 License-File: LICENSE
@@ -101,9 +101,9 @@ target = 'Survived'
 min_stable_attributes = 2
 min_flexible_attributes = 1  # min 1
 min_undesired_support = 1
-min_undesired_confidence = 0.5  # min 0.5
+min_undesired_confidence = 0.5
 min_desired_support = 1
-min_desired_confidence = 0.5  # min 0.5
+min_desired_confidence = 0.5
 undesired_state = '0'
 desired_state = '1'
 # Action Rules Mining

{action_rules-2.0.0 → action_rules-2.0.1}/README.md

@@ -48,9 +48,9 @@ target = 'Survived'
 min_stable_attributes = 2
 min_flexible_attributes = 1  # min 1
 min_undesired_support = 1
-min_undesired_confidence = 0.5  # min 0.5
+min_undesired_confidence = 0.5
 min_desired_support = 1
-min_desired_confidence = 0.5  # min 0.5
+min_desired_confidence = 0.5
 undesired_state = '0'
 desired_state = '1'
 # Action Rules Mining

{action_rules-2.0.0 → action_rules-2.0.1}/pyproject.toml

@@ -1,7 +1,7 @@
 [tool]
 [tool.poetry]
 name = "action-rules"
-version = "2.0.0"
+version = "2.0.1"
 homepage = "https://github.com/lukassykora/action-rules"
 description = "The package for action rules mining using Action-Apriori (Apriori Modified for Action Rules Mining).."
 authors = ["Lukas Sykora <lukas.sykora@vse.cz>"]

{action_rules-2.0.0 → action_rules-2.0.1}/src/action_rules/__init__.py

@@ -17,4 +17,4 @@ __all__ = [
 ]
 __author__ = """Lukas Sykora"""
 __email__ = 'lukas.sykora@vse.cz'
-__version__ = '2.0.0'
+__version__ = '2.0.1'

{action_rules-2.0.0 → action_rules-2.0.1}/tests/test_action_rules.py

@@ -675,3 +675,48 @@ def test_fit_uses_bfs_candidate_expansion(action_rules, monkeypatch):
     )
 
     assert visited_prefixes == [tuple(), ('a',), ('b',), ('a', 'a1')]
+
+
+def test_mining_below_half_confidence():
+    """
+    Test that confidence thresholds below 0.5 are accepted and honoured.
+
+    The package does not enforce a lower bound on
+    ``min_undesired_confidence`` / ``min_desired_confidence``; mining must
+    therefore also return rules whose side confidences lie below 0.5.
+
+    Asserts
+    -------
+    Asserts that rules are mined with thresholds of 0.3 and that at least
+    one discovered rule has an undesired-side confidence below 0.5.
+    """
+    import json
+
+    df = pd.DataFrame(
+        {
+            'A': ['x'] * 10,
+            'F': ['a'] * 5 + ['b'] * 5,
+            'T': ['u', 'u', 'd', 'd', 'd', 'd', 'd', 'd', 'd', 'u'],
+        }
+    )
+    ar = ActionRules(
+        min_stable_attributes=1,
+        min_flexible_attributes=1,
+        min_undesired_support=1,
+        min_undesired_confidence=0.3,
+        min_desired_support=1,
+        min_desired_confidence=0.3,
+        verbose=False,
+    )
+    ar.fit(
+        data=df,
+        stable_attributes=['A'],
+        flexible_attributes=['F'],
+        target='T',
+        target_undesired_state='u',
+        target_desired_state='d',
+    )
+    exported = json.loads(ar.get_rules().get_export_notation())
+    assert len(exported) > 0
+    undesired_confidences = [float(r['confidence of undesired part']) for r in exported]
+    assert any(c < 0.5 for c in undesired_confidences)