def test_seed(self):
seed = 195
dtree1 = dt.DecisionTree(max_features=1, seed=seed)
dtree2 = dt.DecisionTree(max_features=1, seed=seed)
dtree1.fit(X_TRAIN, self.labels_train)
dtree2.fit(X_TRAIN, self.labels_train)
for desc1, desc2 in zip(dtree1.tree.descendents(),
dtree2.tree.descendents()):
assert desc1.split == desc2.split
def setup(self):
self.x_train = X_TRAIN.copy()
self.x_test = X_TEST.copy()
self.x_train[:, 2] = -0.5 + self.labels_train
self.x_test[:, 2] = -0.5 + self.labels_test
self.dtree = dt.DecisionTree()
self.dtree.fit(self.x_train, self.labels_train)
def test_copy(self):
dtree1 = dt.DecisionTree()
dtree1.fit(X_TRAIN, self.labels_train)
dtree2 = dtree1.copy()
assert dtree1.tree is not dtree2.tree
for desc1, desc2 in zip(dtree1.tree.descendents(),
dtree2.tree.descendents()):
assert desc1.split == desc2.split
def setup(self):
self.x_train = X_TRAIN.copy()
self.x_test = X_TEST.copy()
self.x_train[:, 2] = np.abs(
self.x_train[:, 2]) * (-1 + 2 * self.labels_train)
self.x_test[:, 2] = np.abs(
self.x_test[:, 2]) * (-1 + 2 * self.labels_test)
self.dtree = dt.DecisionTree()
self.dtree.fit(self.x_train, self.labels_train)
def test_params(self):
param_dict = {
'min_obs_split': 4,
'max_depth': 2,
'objfunc': metrics.gini,
'max_features': 2,
'seed': 153
}
dtree = dt.DecisionTree(**param_dict)
for key, val in param_dict.items():
assert dtree.params[key] == val
def setup(self):
self.x_train = X_TRAIN.copy()
self.x_test = X_TEST.copy()
self.x_train[:, 2] = (
-0.5 + self.labels_train +
np.logical_and(self.labels_train, self.x_train[:, 2] > 0))
self.x_test[:, 2] = (
-0.5 + self.labels_test +
np.logical_and(self.labels_test, self.x_test[:, 2] > 0))
self.dtree = dt.DecisionTree()
self.dtree.fit(self.x_train, self.labels_train)
def test_params_empty(self):
param_dict = {
'min_obs_split': None,
'max_depth': None,
'objfunc': metrics.gini,
'max_features': None,
'seed': None
}
dtree = dt.DecisionTree(**param_dict)
for key, val in param_dict.items():
if key == 'max_depth':
assert dtree.params[key] == np.inf
else:
assert dtree.params[key] == val
def test_max_features_outofrange2(self):
dtree = dt.DecisionTree(max_depth=1, max_features=-1)
dtree.fit(X_TRAIN, self.labels_train)
def test_max_features_inrange(self):
dtree = dt.DecisionTree(max_depth=1, max_features=2)
dtree.fit(X_TRAIN, self.labels_train)
def test_fit(self):
dtree = dt.DecisionTree()
dtree.fit(X_TRAIN, self.labels_train)
assert dtree.tree is not None
assert dtree.train_err is not None
def setup(self):
self.dtree = dt.DecisionTree()
def setup(self):
self.dtree = dt.DecisionTree()
self.dtree.fit(X_TRAIN, self.labels_train)
def test_print(self):
dtree = dt.DecisionTree()
dtree.fit(X_TRAIN, self.labels_train)
print(dtree)
def test_classify(self):
dtree = dt.DecisionTree()
dtree.fit(X_TRAIN, self.labels_train)
assert np.all(dtree.classify(X_TRAIN) == self.labels_train)
assert np.all(dtree.classify(X_TRAIN[5]) == self.labels_train[5])