def test_create_tree(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert(tree.split[0] == 2)
for i in range(2):
assert tree.children[i].majority == i
assert tree.children[1].split is None
def test_create_tree(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert (tree.split[0] == 2)
for i in range(2):
assert tree.children[i].majority == i
assert tree.children[1].split is None
def test_min_obs_split(self):
min_obs = 5
tree = dt._create_decision_node(X_TRAIN, self.labels_train,
min_obs_split=min_obs)
for desc in tree.descendents():
if len(dt._data_at_node(tree, desc, X_TRAIN)) < min_obs:
assert desc.split is None
def test_min_obs_split(self):
min_obs = 5
tree = dt._create_decision_node(X_TRAIN,
self.labels_train,
min_obs_split=min_obs)
for desc in tree.descendents():
if len(dt._data_at_node(tree, desc, X_TRAIN)) < min_obs:
assert desc.split is None
def test_first_split(self):
tree = dt._create_decision_node(self.x, self.labels_train)
x_sorted = np.sort(self.x[:,2])
zero_ind = np.searchsorted(x_sorted, 0)
assert tree.split[0] == 2
assert (tree.split[1] == x_sorted[zero_ind] or
tree.split[1] == x_sorted[zero_ind-1])
def test_first_split(self):
tree = dt._create_decision_node(self.x, self.labels_train)
x_sorted = np.sort(self.x[:, 2])
zero_ind = np.searchsorted(x_sorted, 0)
assert tree.split[0] == 2
assert (tree.split[1] == x_sorted[zero_ind]
or tree.split[1] == x_sorted[zero_ind - 1])
def test__data_at_node(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert np.all(
self.x[self.x[:,2] < 0] == dt._data_at_node(tree, tree.children[0],
self.x)
)
assert np.all(
self.x[self.x[:,2] > 0] == dt._data_at_node(tree, tree.children[1],
self.x)
)
def test_num_nodes_min_split(self):
tree_minobs0 = dt._create_decision_node(X_TRAIN, self.labels_train)
tree_minobs10 = dt._create_decision_node(X_TRAIN,
self.labels_train,
min_obs_split=10)
assert tree_minobs0.num_nodes() > tree_minobs10.num_nodes()
def test_stumps(self):
tree = dt._create_decision_node(X_TRAIN, self.labels_train)
for stump in tree.stumps():
for i in range(2):
assert (stump.children[i].split is None)
def test_stumps(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert (len(tree.stumps()) == 1)
assert (tree.stumps()[0] is tree)
def test__data_at_node(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert np.all(self.x[self.x[:, 2] < 0] == dt._data_at_node(
tree, tree.children[0], self.x))
assert np.all(self.x[self.x[:, 2] > 0] == dt._data_at_node(
tree, tree.children[1], self.x))
def test_stumps_nodepth(self):
tree = dt._create_decision_node(self.x, self.labels_train, max_depth=0)
assert(len(tree.stumps()) == 0)
def test_tree_max_depth(self):
tree = dt._create_decision_node(self.x, self.labels_train, max_depth=0)
assert tree.split is None
def test_num_nodes_min_split(self):
tree_minobs0 = dt._create_decision_node(X_TRAIN, self.labels_train)
tree_minobs10 = dt._create_decision_node(X_TRAIN, self.labels_train,
min_obs_split=10)
assert tree_minobs0.num_nodes() > tree_minobs10.num_nodes()
def test_classify(self):
tree = dt._create_decision_node(X_TRAIN, self.labels_train)
for obs, label in zip(X_TRAIN, self.labels_train):
assert tree.classify_obs(obs) == label
def test_stumps(self):
tree = dt._create_decision_node(X_TRAIN, self.labels_train)
for stump in tree.stumps():
for i in range(2):
assert(stump.children[i].split is None)
def test_tree_print(self):
tree = dt._create_decision_node(X_TRAIN, self.labels_train)
print(tree)
def test_num_nodes1(self):
tree = dt._create_decision_node(self.x, self.labels_train, max_depth=0)
assert tree.num_nodes() == 1
def test_tree_max_depth(self):
tree = dt._create_decision_node(self.x, self.labels_train, max_depth=0)
assert tree.split is None
def test_stumps(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert(len(tree.stumps()) == 1)
assert(tree.stumps()[0] is tree)
def test_num_nodes(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert tree.num_nodes() == 3
def test_num_nodes(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert tree.num_nodes() == 3
def test_descendents(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert len(tree.descendents()) == 3
assert tree in tree.descendents()
for i in range(2):
assert tree.children[i] in tree.descendents()
def test_num_nodes1(self):
tree = dt._create_decision_node(self.x, self.labels_train, max_depth=0)
assert tree.num_nodes() == 1
def test_classify(self):
tree = dt._create_decision_node(self.x, self.labels_train)
for obs, label in zip(self.x, self.labels_train):
assert tree.classify_obs(obs) == label
def test_tree_print(self):
tree = dt._create_decision_node(X_TRAIN, self.labels_train)
print(tree)
def test_stumps_nodepth(self):
tree = dt._create_decision_node(self.x, self.labels_train, max_depth=0)
assert (len(tree.stumps()) == 0)
def test_descendents(self):
tree = dt._create_decision_node(self.x, self.labels_train)
assert len(tree.descendents()) == 3
assert tree in tree.descendents()
for i in range(2):
assert tree.children[i] in tree.descendents()
