""" Submission tests for Machine Learning Decision Tree Project Author: Nathan Sprague Version: 8/28/2025 """ import time import unittest import numpy as np import sklearn.datasets from sklearn.tree import DecisionTreeClassifier as DecisionTree class MyTestCase(unittest.TestCase): def setUp(self): # These data sets are carefully selected so that there should be # no ties during tree construction. This means that there should # be a unique correct tree for each. self.X2 = np.array([[0.88, 0.39], [0.49, 0.52], [0.68, 0.26], [0.57, 0.51], [0.61, 0.73]]) self.y2 = np.array([0, 1, 1, 1, 0]) self.X2_big = np.array([[0.41, 0.17], [0.45, 0.29], [0.96, 0.46], [0.67, 0.19], [0.76, 0.2], [0.75, 0.59], [0.24, 0.1], [0.82, 0.79], [0.08, 0.16], [0.62, 0.44], [0.22, 0.74], [0.5, 0.48]]) self.y2_big = np.array([0, 0, 0, 0, 0, 1, 1, 0, 1, 1, 1, 0]) self.X3 = np.array([[-2.4, -1.7, 1.2], [-3.6, 4.7, 0.2], [1.9, 2., -1.5], [1.4, -0.9, -0.6], [4.8, -0.7, -1.8], [-1.4, 4.3, -4.9], [-4.7, -2.7, 2.4], [-4., 3.7, -2.7], [-1.6, 3.7, 2.6], [-1.5, -3.1, -0.9], [-2.4, -4.7, 0.6], [4.3, 0.2, 2.]]) self.y3 = np.array([1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 0]) self.X2_3classes = np.array([[0.72, 0.16], [0.18, 0.37], [0.02, 0.53], [0.97, 0.26], [0.38, 0.], [0.61, 0.71], [0.53, 0.2], [0.66, 0.42], [0.78, 0.88], [0.79, 0.26]]) self.y2_3classes = np.array([0, 2, 2, 2, 0, 1, 0, 0, 1, 0]) def full_depth_predict_on_training_helper(self, X, y, noise=0): tree = DecisionTree() tree.fit(X, y) y_test = tree.predict(X + (np.random.random(X.shape) - .5) * noise) np.testing.assert_array_equal(y, y_test) def test_full_depth_predict_on_training_points(self): self.full_depth_predict_on_training_helper(self.X2, self.y2) self.full_depth_predict_on_training_helper(self.X2_big, self.y2_big) self.full_depth_predict_on_training_helper(self.X2_3classes, self.y2_3classes) self.full_depth_predict_on_training_helper(self.X3, self.y3) def test_full_depth_predict_on_perturbed_training_points(self): # Since the training data is all rounded to 2 decimal places, this # amount of perturbation shouldn't be able to push us across a # split boundary. self.full_depth_predict_on_training_helper(self.X2, self.y2, .009) self.full_depth_predict_on_training_helper(self.X2_big, self.y2_big, .009) self.full_depth_predict_on_training_helper(self.X2_3classes, self.y2_3classes, .009) self.full_depth_predict_on_training_helper(self.X3, self.y3, .009) def test_depth_2_tree_predictions(self): tree = DecisionTree(max_depth=2) tree.fit(self.X2_big, self.y2_big) X_test = np.array([[.6, .6], [.4, .8], [.2, .4], [.6, .2]]) y_test = tree.predict(X_test) np.testing.assert_array_equal(np.array([0, 0, 1, 0]), y_test) tree = DecisionTree(max_depth=2) tree.fit(self.X2_3classes, self.y2_3classes) y_test = tree.predict(X_test) np.testing.assert_array_equal(np.array([0, 1, 2, 0]), y_test) def tree_depth_helper(self, X, y, max_depth, expected_depth): tree = DecisionTree(max_depth=max_depth) tree.fit(X, y) self.assertEqual(expected_depth, tree.get_depth()) def test_get_depth_one(self): self.tree_depth_helper(self.X2, self.y2, 1, 1) self.tree_depth_helper(self.X2_big, self.y2_big, 1, 1) self.tree_depth_helper(self.X2_3classes, self.y2_3classes, 1, 1) self.tree_depth_helper(self.X3, self.y3, 1, 1) def test_get_depth_full_depth(self): self.tree_depth_helper(self.X2, self.y2, None, 2) self.tree_depth_helper(self.X2_big, self.y2_big, None, 4) self.tree_depth_helper(self.X2_3classes, self.y2_3classes, None, 3) self.tree_depth_helper(self.X3, self.y3, None, 4) def test_build_tree_efficiency(self): X, y = sklearn.datasets.make_classification(n_samples=1000, n_features=2, n_informative=2, n_redundant=0) tree = DecisionTree() start = time.time() tree.fit(X, y) elapsed = time.time() - start self.assertTrue(elapsed < 10.0, "too slow: {:.2f}(s)".format(elapsed)) def test_score_full_tree(self): X, y = sklearn.datasets.make_classification(n_samples=1000, n_features=2, n_informative=2, n_redundant=0) tree = DecisionTree() tree.fit(X, y) score = tree.score(X, y) # Since we fit on the training data, we should get a perfect score self.assertAlmostEqual(score, 1.0, places=4) def test_score_datasets_from_setup(self): # Test the score method on the datasets defined in setUp tree = DecisionTree() tree.fit(self.X2, self.y2) score = tree.score(self.X2, self.y2) self.assertAlmostEqual(score, 1.0, places=4) # create a new y for a non 1.0 score y2_modified = np.array([0, 1, 1, 0, 0]) score = tree.score(self.X2, y2_modified) self.assertAlmostEqual(score, 0.8, places=4) # create a new y for a non 1.0 score y2_3classes_modified = np.array([0, 2, 1, 2, 0, 1, 0, 0, 1, 0]) score = tree.score(self.X2_3classes, y2_3classes_modified) self.assertAlmostEqual(score, 0.2, places=4) tree.fit(self.X3, self.y3) score = tree.score(self.X3, self.y3) self.assertAlmostEqual(score, 1.0, places=4) # create a new y for a non 1.0 score y3_modified = np.array([1, 0, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1]) score = tree.score(self.X3, y3_modified) self.assertAlmostEqual(score, 0.9166666667, places=4) def test_feature_importances_reasonable(self): # Test feature importances on a dataset X, y = sklearn.datasets.make_classification(n_samples=1000, n_features=5, n_informative=3, n_redundant=0) tree = DecisionTree() tree.fit(X, y) importances = tree.feature_importances_ # Check that the importances are non-negative and sum to 1 self.assertTrue(np.all(importances >= 0)) self.assertAlmostEqual(np.sum(importances), 1.0, places=4) def test_feature_importances_on_setup_data(self): # Test feature importances on the datasets defined in setUp tree = DecisionTree() tree.fit(self.X2, self.y2) importances = tree.feature_importances_ np.testing.assert_array_almost_equal(importances, np.array([0.44444444, 0.55555556]), decimal=4) tree.fit(self.X2_big, self.y2_big) importances = tree.feature_importances_ np.testing.assert_array_almost_equal(importances, np.array([0.87809524, 0.12190476]), decimal=4) tree.fit(self.X2_3classes, self.y2_3classes) importances = tree.feature_importances_ np.testing.assert_array_almost_equal(importances, np.array([0.60483871, 0.39516129]), decimal=4) tree.fit(self.X3, self.y3) importances = tree.feature_importances_ np.testing.assert_array_almost_equal(importances, np.array([0.28, 0.30857143, 0.41142857]), decimal=4) if __name__ == '__main__': unittest.main()