2022吴恩达机器学习专项课程C3作业第一周

import numpy as np import random def select_threshold_test(target): p_val = np.array([i / 100 for i in range(30)]) y_val = np.array([1] * 5 + [0] * 25) best_epsilon, best_F1 = target(y_val, p_val) assert np.isclose(best_epsilon, 0.04, atol=0.3 / 1000), f"Wrong best_epsilon. Expected: {0.04} got: {best_epsilon}" assert best_F1 == 1, f"Wrong best_F1. Expected: 1 got: {best_F1}" y_val = np.array([1] * 5 + [0] * 25) y_val[2] = 0 # Introduce noise best_epsilon, best_F1 = target(y_val, p_val) assert np.isclose(best_epsilon, 0.04, atol=0.3 / 1000), f"Wrong best_epsilon. Expected: {0.04} got: {best_epsilon}" assert np.isclose(best_F1, 0.8888888), f"Wrong best_F1. Expected: 0.8888888 got: {best_F1}" p_val = np.array([i / 1000 for i in range(50)]) y_val = np.array([1] * 8 + [0] * 42) y_val[5] = 0 index = [*range(50)] random.shuffle(index) p_val = p_val[index] y_val = y_val[index] best_epsilon, best_F1 = target(y_val, p_val) assert np.isclose(best_epsilon, 0.007, atol=0.05 / 1000), f"Wrong best_epsilon. Expected: {0.0070070} got: {best_epsilon}" assert np.isclose(best_F1, 0.933333333), f"Wrong best_F1. Expected: 0.933333333 got: {best_F1}" print("\033[92mAll tests passed!") def estimate_gaussian_test(target): np.random.seed(273) X = np.array([[1, 1, 1], [2, 2, 2], [3, 3, 3]]).T mu, var = target(X) assert type(mu) == np.ndarray, f"Wrong type for mu. Expected: {np.ndarray} got: {type(mu)}" assert type(var) == np.ndarray, f"Wrong type for mu. Expected: {np.ndarray} got: {type(var)}" assert mu.shape == (X.shape[1],), f"Wrong shape for mu. Expected: {(X.shape[1],)} got: {mu.shape}" assert type(var) == np.ndarray, f"Wrong shape for mu. Expected: {(X.shape[1],)} got: {var.shape}" assert np.allclose(mu, [1., 2., 3.]), f"Wrong value for mu. Expected: {[1, 2, 3]} got: {mu}" assert np.allclose(var, [0., 0., 0.]), f"Wrong value for mu. Expected: {[0, 0, 0]} got: {var}" X = np.array([[1, 2, 3], [2, 4, 6], [3, 6, 9]]).T mu, var = target(X) assert type(mu) == np.ndarray, f"Wrong type for mu. Expected: {np.ndarray} got: {type(mu)}" assert type(var) == np.ndarray, f"Wrong type for mu. Expected: {np.ndarray} got: {type(var)}" assert mu.shape == (X.shape[1],), f"Wrong shape for mu. Expected: {(X.shape[1],)} got: {mu.shape}" assert type(var) == np.ndarray, f"Wrong shape for mu. Expected: {(X.shape[1],)} got: {var.shape}" assert np.allclose(mu, [2., 4., 6.]), f"Wrong value for mu. Expected: {[2., 4., 6.]} got: {mu}" assert np.allclose(var, [2. / 3, 8. / 3., 18. / 3.]), f"Wrong value for mu. Expected: {[2. / 3, 8. / 3., 18. / 3.]} got: {var}" m = 500 X = np.array([np.random.normal(0, 1, m), np.random.normal(1, 2, m), np.random.normal(3, 1.5, m)]).T mu, var = target(X) assert type(mu) == np.ndarray, f"Wrong type for mu. Expected: {np.ndarray} got: {type(mu)}" assert type(var) == np.ndarray, f"Wrong type for mu. Expected: {np.ndarray} got: {type(var)}" assert mu.shape == (X.shape[1],), f"Wrong shape for mu. Expected: {(X.shape[1],)} got: {mu.shape}" assert type(var) == np.ndarray, f"Wrong shape for mu. Expected: {(X.shape[1],)} got: {var.shape}" assert np.allclose(mu, [0., 1., 3.], atol=0.2), f"Wrong value for mu. Expected: {[0, 1, 3]} got: {mu}" assert np.allclose(var, np.square([1., 2., 1.5]), atol=0.2), f"Wrong value for mu. Expected: {np.square([1., 2., 1.5])} got: {var}" print("\033[92mAll tests passed!")