def test_apply(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
leaves = tree.apply(iris_X)
assert isinstance(leaves, np.ndarray)
assert np.all(leaves > 0)
assert len(leaves) == len(iris_X)
def test_importance(self, iris_X, iris_y, importance,
scale_permutation_importance, local_importance):
tree = RangerTreeClassifier(
importance=importance,
scale_permutation_importance=scale_permutation_importance,
local_importance=local_importance,
)
if importance not in [
"none", "impurity", "impurity_corrected", "permutation"
]:
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
return
tree.fit(iris_X, iris_y)
if importance == "none":
assert tree.importance_mode_ == 0
elif importance == "impurity":
assert tree.importance_mode_ == 1
elif importance == "impurity_corrected":
assert tree.importance_mode_ == 5
elif importance == "permutation":
if local_importance:
assert tree.importance_mode_ == 6
elif scale_permutation_importance:
assert tree.importance_mode_ == 2
else:
assert tree.importance_mode_ == 3
def test_verbose(self, iris_X, iris_y, verbose, capfd):
tree = RangerTreeClassifier(verbose=verbose)
tree.fit(iris_X, iris_y)
captured = capfd.readouterr()
if verbose:
assert len(captured.out) > 0
else:
assert len(captured.out) == 0
def test_sample_fraction_replace(self, iris_X, iris_y, replace):
tree = RangerTreeClassifier(replace=replace)
tree.fit(iris_X, iris_y)
if replace:
assert tree.sample_fraction_ == [1.0]
else:
assert tree.sample_fraction_ == [0.632]
def test_serialize(self, iris_X, iris_y):
tf = tempfile.TemporaryFile()
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
pickle.dump(tree, tf)
tf.seek(0)
new_tree = pickle.load(tf)
pred = new_tree.predict(iris_X)
assert len(pred) == iris_X.shape[0]
def test_predict_log_proba(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
pred = tree.predict_log_proba(iris_X)
assert len(pred) == iris_X.shape[0]
# test with single record
iris_X_record = iris_X[0:1, :]
pred = tree.predict_log_proba(iris_X_record)
assert len(pred) == 1
def test_fit(self, iris_X, iris_y):
tree = RangerTreeClassifier()
with pytest.raises(NotFittedError):
check_is_fitted(tree)
tree.fit(iris_X, iris_y)
check_is_fitted(tree)
assert hasattr(tree, "classes_")
assert hasattr(tree, "n_classes_")
assert hasattr(tree, "ranger_forest_")
assert hasattr(tree, "ranger_class_order_")
assert hasattr(tree, "n_features_in_")
def test_regularization(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
assert tree.regularization_factor_ == []
assert not tree.use_regularization_factor_
# vector must be between 0 and 1 and length matching feature num
for r in [[1.1], [-0.1], [1, 1]]:
tree = RangerTreeClassifier(regularization_factor=r)
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
# vector of ones isn't applied
tree = RangerTreeClassifier(regularization_factor=[1] *
iris_X.shape[1])
tree.fit(iris_X, iris_y)
assert tree.regularization_factor_ == []
assert not tree.use_regularization_factor_
# regularization vector is used
reg = [0.5]
tree = RangerTreeClassifier(regularization_factor=reg)
tree.fit(iris_X, iris_y)
assert tree.regularization_factor_ == reg
assert tree.use_regularization_factor_
def test_sample_fraction(self, iris_X, iris_y):
tree = RangerTreeClassifier(sample_fraction=[0.69])
tree.fit(iris_X, iris_y)
assert tree.sample_fraction_ == [0.69]
tree = RangerTreeClassifier(sample_fraction=0.69)
tree.fit(iris_X, iris_y)
assert tree.sample_fraction_ == [0.69]
# test with single record
iris_X_record = iris_X[0:1, :]
pred = tree.predict(iris_X_record)
assert len(pred) == 1
pred = tree.predict_proba(iris_X_record)
assert len(pred) == 1
pred = tree.predict_log_proba(iris_X_record)
assert len(pred) == 1
def test_tree_interface(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
# access attributes the way we would expect to in sklearn
tree_ = tree.tree_
children_left = tree_.children_left
children_right = tree_.children_right
feature = tree_.feature
threshold = tree_.threshold
max_depth = tree_.max_depth
n_node_samples = tree_.n_node_samples
weighted_n_node_samples = tree_.weighted_n_node_samples
node_count = tree_.node_count
capacity = tree_.capacity
n_outputs = tree_.n_outputs
n_classes = tree_.n_classes
value = tree_.value
assert value.shape == (node_count, 1, len(np.unique(iris_y)))
def test_accuracy(self, iris_X, iris_y):
X_train, X_test, y_train, y_test = train_test_split(iris_X,
iris_y,
test_size=0.33,
random_state=42)
# train and test a random forest classifier
rf = RandomForestClassifier()
rf.fit(X_train, y_train)
y_pred_rf = rf.predict(X_test)
rf_acc = accuracy_score(y_test, y_pred_rf)
# train and test a ranger classifier
ra = RangerTreeClassifier()
ra.fit(X_train, y_train)
y_pred_ra = ra.predict(X_test)
ranger_acc = accuracy_score(y_test, y_pred_ra)
# the accuracy should be good
assert rf_acc > 0.9
assert ranger_acc > 0.9
def test_class_weights(self, iris_X, iris_y):
X_train, X_test, y_train, y_test = train_test_split(iris_X,
iris_y,
test_size=0.5,
random_state=42)
tree = RangerTreeClassifier()
weights = {
0: 0.7,
1: 0.2,
2: 0.1,
}
tree.fit(X_train, y_train, class_weights=weights)
tree.predict(X_test)
tree = RangerTreeClassifier()
m = {0: "a", 1: "b", 2: "c"}
y_train_str = [m.get(v) for v in y_train]
weights = {
"a": 0.7,
"b": 0.2,
"c": 0.1,
}
tree.fit(X_train, y_train_str, class_weights=weights)
tree.predict(X_test)
weights = {
0: 0.7,
}
with pytest.raises(ValueError):
tree.fit(X_train, y_train, class_weights=weights)
def test_categorical_features(self, iris_X, iris_y,
respect_categorical_features):
# add a categorical feature
categorical_col = np.atleast_2d(
np.array([random.choice([0, 1]) for _ in range(iris_X.shape[0])]))
iris_X_c = np.hstack((iris_X, categorical_col.transpose()))
categorical_features = [iris_X.shape[1]]
tree = RangerTreeClassifier(
respect_categorical_features=respect_categorical_features, )
if respect_categorical_features not in [
"partition", "ignore", "order"
]:
with pytest.raises(ValueError):
tree.fit(iris_X_c,
iris_y,
categorical_features=categorical_features)
return
tree.fit(iris_X_c, iris_y, categorical_features=categorical_features)
tree.predict(iris_X_c)
def test_split_rule(self, iris_X, iris_y, split_rule):
tree = RangerTreeClassifier(split_rule=split_rule)
assert tree.criterion == split_rule
if split_rule not in ["gini", "extratrees", "hellinger"]:
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
return
# hellinger can only be used in binary classification
if split_rule == "hellinger":
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
iris_2 = [0 if v == 2 else v for v in iris_y]
tree.fit(iris_X, iris_2)
if split_rule == "gini":
assert tree.split_rule_ == 1
elif split_rule == "extratrees":
assert tree.split_rule_ == 5
if split_rule == "hellinger":
assert tree.split_rule_ == 7
if split_rule == "extratrees":
tree = RangerTreeClassifier(
split_rule=split_rule,
respect_categorical_features="partition",
save_memory=True,
)
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
else:
tree = RangerTreeClassifier(split_rule=split_rule,
num_random_splits=2)
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
def test_split_select_weights(self, iris_X, iris_y):
n_trees = 1
weights = [0.1] * iris_X.shape[1]
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y, split_select_weights=weights)
weights = [0.1] * (iris_X.shape[1] - 1)
tree = RangerTreeClassifier()
with pytest.raises(RuntimeError):
tree.fit(iris_X, iris_y, split_select_weights=weights)
weights = [[0.1] * (iris_X.shape[1])] * n_trees
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y, split_select_weights=weights)
weights = [[0.1] * (iris_X.shape[1])] * (n_trees + 1)
tree = RangerTreeClassifier()
with pytest.raises(RuntimeError):
tree.fit(iris_X, iris_y, split_select_weights=weights)
def test_inbag(self, iris_X, iris_y):
inbag = [[1, 2, 3]]
tree = RangerTreeClassifier(inbag=inbag)
tree.fit(iris_X, iris_y)
# can't use inbag with sample weight
tree = RangerTreeClassifier(inbag=inbag)
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y, sample_weight=[1] * len(iris_y))
# can't use class sampling and inbag
tree = RangerTreeClassifier(inbag=inbag, sample_fraction=[1, 1])
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
def test_mtry(self, iris_X, iris_y, mtry):
tree = RangerTreeClassifier(mtry=mtry)
if callable(mtry) and mtry(5) > 5:
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
return
elif not callable(mtry) and (mtry < 0 or mtry > iris_X.shape[0]):
with pytest.raises(ValueError):
tree.fit(iris_X, iris_y)
return
tree.fit(iris_X, iris_y)
if callable(mtry):
assert tree.mtry_ == mtry(iris_X.shape[1])
else:
assert tree.mtry_ == mtry
def test_clone(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
clone(tree)
def test_decision_path(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
paths = tree.decision_path(iris_X)
assert isinstance(paths, csr_matrix)
assert paths.shape[0] == len(iris_X)
def test_always_split_features(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y, always_split_features=[0])
# feature 0 is in every tree split
for tree in tree.ranger_forest_["forest"]["split_var_ids"]:
assert 0 in tree
def test_get_n_leaves(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
leaves = tree.get_n_leaves()
assert isinstance(leaves, int)
assert np.all(leaves > 0)
def test_get_depth(self, iris_X, iris_y):
tree = RangerTreeClassifier()
tree.fit(iris_X, iris_y)
depth = tree.get_depth()
assert isinstance(depth, int)
assert depth > 0