update

Author: wkd-lidashuang

ex1/machine-learning-ex1.py

@@ -5,9 +5,6 @@
 
 
 def plot_data(data):
-    # print(data.head())
-    # print(data.describe())
-    # print(data)
     data.plot(kind='scatter', x='Population', y='Profit', figsize=(8, 5))
     plt.xlabel('Population of City in 10,000s')
     plt.ylabel('Profit in $10,000s')

ex2/machine-learning-ex2.py

@@ -4,8 +4,7 @@
 import scipy.optimize as opt
 
 
-def visualizing_data(data):
-    print(data.head())
+def plotData(data):
     pos = data[data.admitted.isin(['1'])]
     neg = data[data.admitted.isin(['0'])]
     plt.scatter(pos['exam1'], pos['exam2'], marker='+', c='k', label='Admitted')
@@ -17,12 +16,24 @@ def visualizing_data(data):
     plt.show()
 
 
+def plotDecisionBoundary(theta, data):
+    x1 = np.arange(100, step=0.1)
+    x2 = -(theta[0] + x1 * theta[1]) / theta[2]
+    pos = data[data.admitted.isin(['1'])]
+    neg = data[data.admitted.isin(['0'])]
+    plt.scatter(pos['exam1'], pos['exam2'], marker='+', c='k', label='Admitted')
+    plt.scatter(neg['exam1'], neg['exam2'], marker='o', c='y', label='Not Admitted')
+    plt.plot(x1, x2)
+    plt.axis([30, 100, 30, 100])
+    plt.xlabel('Exam 1 Score')
+    plt.ylabel('Exam 2 Score')
+    plt.legend(loc=1, prop={'size': 9})
+    plt.title('Decision Boundary')
+
+
 def visualizing_data2(data):
     print(data.head())
-    pos = data[data.Accepted.isin(['1'])]
-    neg = data[data.Accepted.isin(['0'])]
-    plt.scatter(pos['Test 1'], pos['Test 2'], marker='+', c='k', label='y=1')
-    plt.scatter(neg['Test 1'], neg['Test 2'], marker='o', c='y', label='y=0')
+
     # 设置横纵坐标名
     plt.xlabel('Microchip Test 1')
     plt.ylabel('Microchip Test 2')
@@ -49,26 +60,6 @@ def predict(theta, x):
     return [1 if x >= 0.5 else 0 for x in probability]
 
 
-def evaluating_logistic(final_theta, x, y, data):
-    predictions = predict(final_theta, x)
-    correct = [1 if a == b else 0 for (a, b) in zip(predictions, y)]
-    accuracy = sum(correct) / len(x)
-    print(accuracy)
-    x1 = np.arange(100, step=0.1)
-    x2 = -(final_theta[0] + x1 * final_theta[1]) / final_theta[2]
-    pos = data[data.admitted.isin(['1'])]
-    neg = data[data.admitted.isin(['0'])]
-    plt.scatter(pos['exam1'], pos['exam2'], marker='+', c='k', label='Admitted')
-    plt.scatter(neg['exam1'], neg['exam2'], marker='o', c='y', label='Not Admitted')
-    plt.plot(x1, x2)
-    plt.axis([30, 100, 30, 100])
-    plt.xlabel('Exam 1 Score')
-    plt.ylabel('Exam 2 Score')
-    plt.legend(loc=1, prop={'size': 9})
-    plt.title('Decision Boundary')
-    plt.show()
-
-
 def feature_mapping(x1, x2, power):
     data = {}
     for i in np.arange(power + 1):
@@ -90,35 +81,91 @@ def gradientReg(theta, X, y, l=1):
     return gradient(theta, X, y) + reg
 
 
-def main():
-    # path = 'ex2data1.txt'
-    # data = pd.read_csv(path, names=['exam1', 'exam2', 'admitted'])
-    # # visualizing_data(data)
-    # data.insert(0, 'Ones', 1)
-    # x = data.iloc[:, :-1].values
-    # y = data.iloc[:, -1].values
-    # theta = np.zeros(x.shape[1])
-    # print(cost(theta, x, y))
-    # print(gradient(theta, x, y))
-    # result = opt.fmin_tnc(func=cost, x0=theta, fprime=gradient, args=(x, y))
-    # final_theta = result[0]
-    # evaluating_logistic(final_theta, x, y, data)
-    data2 = pd.read_csv('ex2data2.txt', names=['Test 1', 'Test 2', 'Accepted'])
-    # visualizing_data2(data2)
-    x1 = data2['Test 1'].values
-    x2 = data2['Test 2'].values
-    _data2 = feature_mapping(x1, x2, power=6)
-    x = _data2.values
-    y = data2['Accepted'].values
-    theta = np.zeros(x.shape[1])
-    print(costReg(theta, x, y, lam=1))
-    result2 = opt.fmin_tnc(func=costReg, x0=theta, fprime=gradientReg, args=(x, y, 2))
-    final_theta = result2[0]
-    predictions = predict(final_theta, x)
-    correct = [1 if a == b else 0 for (a, b) in zip(predictions, y)]
-    accuracy = sum(correct) / len(correct)
-    print(accuracy)
-
-
-if __name__ == '__main__':
-    main()
+# ==================== Part 1: Plotting ====================
+print('Plotting data with + indicating (y = 1) examples and o indicating (y = 0) examples.')
+path = 'ex2data1.txt'
+data = pd.read_csv(path, names=['exam1', 'exam2', 'admitted'])
+plotData(data)
+
+data.insert(0, 'Ones', 1)
+X = data.iloc[:, :-1].values
+y = data.iloc[:, -1].values
+input("Program paused. Press Enter to continue...")
+
+# # ============ Part 2: Compute Cost and Gradient ============
+initial_theta = np.zeros(X.shape[1])
+J = cost(initial_theta, X, y)
+print('Cost at initial theta (zeros): %f' % J)
+grad = gradient(initial_theta, X, y)
+print('Gradient at initial theta (zeros): ' + str(grad))
+
+input("Program paused. Press Enter to continue...")
+
+# ============= Part 3: Optimizing using scipy  =============
+result = opt.fmin_tnc(func=cost, x0=initial_theta, fprime=gradient, args=(X, y))
+theta = result[0]
+# Plot Boundary
+plotDecisionBoundary(theta, data)
+plt.show()
+input("Program paused. Press Enter to continue...")
+
+#  ============== Part 4: Predict and Accuracies ==============
+prob = sigmoid(np.array([1, 45, 85]).dot(theta))
+print('For a student with scores 45 and 85, we predict an admission probability of %f' % prob)
+
+# Compute accuracy on our training set
+p = predict(theta, X)
+correct = [1 if a == b else 0 for (a, b) in zip(p, y)]
+accuracy = sum(correct) / len(correct)
+print('Train Accuracy: %f' % accuracy)
+
+input("Program paused. Press Enter to continue...")
+
+# =========== Part 5: Regularized Logistic Regression ============
+
+data2 = pd.read_csv('ex2data2.txt', names=['Test 1', 'Test 2', 'Accepted'])
+
+x1 = data2['Test 1'].values
+x2 = data2['Test 2'].values
+_data2 = feature_mapping(x1, x2, power=6)
+X = _data2.values
+y = data2['Accepted'].values
+initial_theta = np.zeros(X.shape[1])
+
+J = costReg(initial_theta, X, y, lam=1)
+print('Cost at initial theta (zeros): %f' % J)
+result2 = opt.fmin_tnc(func=costReg, x0=initial_theta, fprime=gradientReg, args=(X, y, 2))
+theta = result2[0]
+predictions = predict(theta, X)
+correct = [1 if a == b else 0 for (a, b) in zip(predictions, y)]
+accuracy = sum(correct) / len(correct)
+print(accuracy)
+
+input("Program paused. Press Enter to continue...")
+
+
+# ============= Part 3: Optional Exercises =============
+
+# Plot Boundary
+def plotBoundary(theta, data2, Lambda):
+    plt.title(r'$\lambda$ = ' + str(Lambda))
+    x = np.linspace(-1, 1.5, 250)
+    xx, yy = np.meshgrid(x, x)
+    z = feature_mapping(xx.ravel(), yy.ravel(), power=6).values
+    z = z @ theta
+    z = z.reshape(xx.shape)
+    pos = data2[data2.Accepted.isin(['1'])]
+    neg = data2[data2.Accepted.isin(['0'])]
+    plt.scatter(pos['Test 1'], pos['Test 2'], marker='+', c='k', label='y=1')
+    plt.scatter(neg['Test 1'], neg['Test 2'], marker='o', c='y', label='y=0')
+    plt.contour(xx, yy, z, 0)
+    plt.ylim(-.8, 1.2)
+    plt.show()
+
+
+for Lambda in np.arange(0.0, 10.1, 1.0):
+    result2 = opt.fmin_tnc(func=costReg, x0=initial_theta, fprime=gradientReg, args=(X, y, Lambda))
+    theta = result2[0]
+    print('lambda = ' + str(Lambda))
+    print('theta:', ["%0.4f" % i for i in theta])
+    plotBoundary(theta, data2, Lambda)

ex3/machine-learning-ex3.py

@@ -12,15 +12,7 @@ def load_data(path):
     return x, y
 
 
-def plot_an_image(X, y):
-    pick_one = np.random.randint(0, 5000)
-    image = X[pick_one, :]
-    plt.matshow(image.reshape((20, 20)))
-    plt.show()
-    print('this should be {}'.format(y[pick_one]))
-
-
-def plot_100_image(X):
+def display_data(X):
     """
     随机画100个数字
     """
@@ -31,7 +23,7 @@ def plot_100_image(X):
 
     for row in range(10):
         for column in range(10):
-            ax_array[row, column].matshow(sample_images[10 * row + column].reshape((20, 20)))
+            ax_array[row, column].matshow(sample_images[10 * row + column].reshape((20, 20)), cmap='gray')
     plt.xticks([])
     plt.yticks([])
     plt.show()
@@ -70,12 +62,12 @@ def regularized_gradient(theta, X, y, l):
     return first + reg
 
 
-def one_vs_all(X, y, l, K):
+def one_vs_all(X, y, Lambda, K):
     """generalized logistic regression
     args:
         X: feature matrix, (m, n+1) # with incercept x0=1
         y: target vector, (m, )
-        l: lambda constant for regularization
+        Lambda: lambda constant for regularization
         K: numbel of labels
     return: trained parameters
     """
@@ -85,10 +77,9 @@ def one_vs_all(X, y, l, K):
         theta = np.zeros(X.shape[1])  # (401,)
         y_i = np.array([1 if label == i else 0 for label in y])
 
-        ret = minimize(fun=regularized_cost, x0=theta, args=(X, y_i, l), method='TNC',
+        ret = minimize(fun=regularized_cost, x0=theta, args=(X, y_i, Lambda), method='TNC',
                        jac=regularized_gradient, options={'disp': True})
         all_theta[i - 1, :] = ret.x
-    print(all_theta.shape)
     return all_theta
 
 
@@ -112,11 +103,9 @@ def load_weight(path):
 def neural_networks():
     # Neural Networks
     theta1, theta2 = load_weight('ex3weights.mat')
-    print(theta1.shape, theta2.shape)
     X, y = load_data('ex3data1.mat')
     y = y.flatten()
     X = np.insert(X, 0, 1, axis=1)  # intercept
-    print(X.shape)
     a1 = X
     z2 = a1 @ theta1.T
     a2 = np.insert(sigmoid(z2), 0, 1, axis=1)
@@ -127,18 +116,27 @@ def neural_networks():
     print('accuracy = {0}%'.format(accuracy * 100))  # accuracy = 97.52%
 
 
-def main():
-    x, y = load_data('ex3data1.mat')
-    # plot_an_image(x, y)
-    # plot_100_image(x)
-    x = np.insert(x, 0, 1, axis=1)  # (5000, 401)
-    y = y.flatten()  # 这里消除了一个维度，方便后面的计算 or .reshape(-1) （5000，）
-    all_theta = one_vs_all(x, y, 1, 10)  # 每一行是一个分类器的一组参数
-    y_pred = predict_all(x, all_theta)
-    accuracy = np.mean(y_pred == y)
-    print('accuracy = {0}%'.format(accuracy * 100))
+## =========== Part 1: Loading and Visualizing Data =============
+print('Loading and Visualizing Data ...')
+X, y = load_data('ex3data1.mat')
+
+# Randomly select 100 data points to display
+display_data(X)
+
+input("Program paused. Press Enter to continue...")
+
+## ============ Part 2: Vectorize Logistic Regression ============
+print('Training One-vs-All Logistic Regression...')
+Lambda = 1
+X = np.insert(X, 0, 1, axis=1)  # (5000, 401)
+y = y.flatten()  # 这里消除了一个维度，方便后面的计算 or .reshape(-1) （5000，）
+all_theta = one_vs_all(X, y, Lambda, 10)
+input("Program paused. Press Enter to continue...")
 
+## ================ Part 3: Predict for One-Vs-All ================
+y_pred = predict_all(X, all_theta)
+accuracy = np.mean(y_pred == y)
+print('accuracy = {0}%'.format(accuracy * 100))
 
-if __name__ == '__main__':
-    # main()
-    neural_networks()
+## ================Neural Networks
+neural_networks()