def test_feature_importance_attribute_exists_for_elasticnet(wandb_init_run):
X, y = make_regression(n_features=2, random_state=42)
two_features = ['a', 'b']
model = ElasticNet(random_state=42)
model.fit(X, y)
result = plot_feature_importances(model, feature_names=two_features)
assert isinstance(result, wandb.viz.Visualize)
def test_feature_importance_attribute_does_not_exists(
wandb_init_run, dummy_model, dummy_data
):
dummy_model.fit(*dummy_data, epochs=2, batch_size=36, callbacks=[WandbCallback()])
dummy_features = []
result = plot_feature_importances(dummy_model, feature_names=dummy_features)
assert result is None
def test_feature_importance_attribute_exists_for_random_forest(wandb_init_run):
X, y = make_hastie_10_2(random_state=0)
model = GradientBoostingClassifier(
n_estimators=100, learning_rate=1.0, max_depth=1, random_state=42
)
ten_features = ["a", "b", "c", "d", "e", "f", "g", "h", "i", "k"]
model.fit(X, y)
result = plot_feature_importances(model, feature_names=ten_features)
assert isinstance(result, wandb.viz.Visualize)
def test_feature_importance_attribute_exists_for_random_forest(wandb_init_run):
X, y = make_hastie_10_2(random_state=0)
model = GradientBoostingClassifier(n_estimators=100,
learning_rate=1.0,
max_depth=1,
random_state=42)
ten_features = ['a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'k']
model.fit(X, y)
result = plot_feature_importances(model, feature_names=ten_features)
assert isinstance(result, wandb.viz.Visualize)
def test_feature_importance_attribute_type(wandb_init_run):
X, y = make_regression(n_features=2, random_state=42)
model = ElasticNet(random_state=42)
model.fit(X, y)
# input np.array instead of list
two_features = np.array(["a", "b"])
result = None
try:
result = plot_feature_importances(model, feature_names=two_features)
except ValueError:
pytest.fail('incorrect plot_feature_importances arguments: ValueError')
assert isinstance(result, wandb.viz.Visualize)