def test_fit(self):
thresholder = SimpleThresholder(self.rules, "or")
assert thresholder.feature_importances_ is None
thresholder.fit(x=self.data["X_train"], y=self.data["y_train"])
np.testing.assert_array_equal(
thresholder.feature_importances_, np.array([1, 1, 0, 0])
)
def test__convert_string_rule_to_dict(self):
thresholder = SimpleThresholder(self.rules, "or")
results = thresholder._convert_string_rule_to_dict(self.rules[0])
expected_results = {
"feature_name": "x1",
"operator": "gt",
"threshold": 0
}
assert results == expected_results
def test__convert_string_rule_to_dict(self):
thresholder = SimpleThresholder(self.rules, 'or')
results = thresholder._convert_string_rule_to_dict(self.rules[0])
expected_results = {
'feature_name': 'x1',
'operator': 'gt',
'threshold': 0
}
assert results == expected_results
def test_predict_proba(self):
for logical_operator in ["and", "or"]:
thresholder = SimpleThresholder(self.rules, logical_operator)
thresholder.fit(x=self.data["X_train"], y=self.data["y_train"])
results = thresholder.predict_proba(self.data["X_test"])
if logical_operator == "and":
expected_results = np.array([[0, 1, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0,
0]]).transpose()
elif logical_operator == "or":
expected_results = np.array([[1, 1, 1, 1, 1, 1, 0, 1],
[1, 1, 1, 1, 1, 1, 0,
1]]).transpose()
np.testing.assert_array_equal(results, expected_results)
def test_predict_proba(self):
for logical_operator in ['and', 'or']:
thresholder = SimpleThresholder(self.rules, logical_operator)
thresholder.fit(x=self.data['X_train'], y=self.data['y_train'])
results = thresholder.predict_proba(self.data['X_test'])
if logical_operator == 'and':
expected_results = np.array([[0, 1, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0,
0]]).transpose()
elif logical_operator == 'or':
expected_results = np.array([[1, 1, 1, 1, 1, 1, 0, 1],
[1, 1, 1, 1, 1, 1, 0,
1]]).transpose()
np.testing.assert_array_equal(results, expected_results)
def test_rule_with_unavailable_feature_raises_error(self):
rules = self.rules
rules.append("x5 == 3")
thresholder = SimpleThresholder(rules, "or")
with self.assertRaises(BaselineFeatureNotInMatrix):
thresholder.fit(x=self.data["X_train"], y=self.data["y_train"])
def test_all_feature_names_property(self):
thresholder = SimpleThresholder(self.rules, "or")
assert thresholder.all_feature_names == ["x1", "x2"]
def test_rule_with_unavailable_feature_raises_error(self):
rules = self.rules
rules.append('x5 == 3')
thresholder = SimpleThresholder(rules, 'or')
with self.assertRaises(BaselineFeatureNotInMatrix):
thresholder.fit(x=self.data['X_train'], y=self.data['y_train'])
def test_fit(self):
thresholder = SimpleThresholder(self.rules, 'or')
assert thresholder.feature_importances_ == None
thresholder.fit(x=self.data['X_train'], y=self.data['y_train'])
np.testing.assert_array_equal(thresholder.feature_importances_,
np.array([1, 1, 0, 0]))
def test_all_feature_names_property(self):
thresholder = SimpleThresholder(self.rules, 'or')
assert thresholder.all_feature_names == ['x1', 'x2']
