[GitHub]Python Machine Learning

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

Python Machine Learning

本帖隐藏的内容

https://github.com/rasbt/python-machine-learning-book

What you can expect are 400 pages rich in useful material just about everything you need to know to get started with machine learning ... from theory to the actual code that you can directly put into action! This is not yet just another "this is how scikit-learn works" book. I aim to explain all the underlying concepts, tell you everything you need to know in terms of best practices and caveats, and we will put those concepts into action mainly using NumPy, scikit-learn, and Theano.

You are not sure if this book is for you? Please checkout the excerpts from the Foreword and Preface, or take a look at the FAQ section for further information.

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏2 回帖

1. First steps with scikit-learn
   
2. Loading the Iris dataset from scikit-learn. Here, the third column represents the petal length, and the fourth column the petal width of the flower samples. The classes are already converted to integer labels where 0=Iris-Setosa, 1=Iris-Versicolor, 2=Iris-Virginica.
   
3. In [4]:
   
4. from sklearn import datasets
   
5. import numpy as np
   
6. 
   
7. iris = datasets.load_iris()
   
8. X = iris.data[:, [2, 3]]
   
9. y = iris.target
   
10. 
    
11. print('Class labels:', np.unique(y))
    
12. 
    
13. 
    
14. In [5]:
    
15. if Version(sklearn_version) < '0.18':
    
16.     from sklearn.cross_validation import train_test_split
    
17. else:
    
18.     from sklearn.model_selection import train_test_split
    
19. 
    
20. X_train, X_test, y_train, y_test = train_test_split(
    
21.     X, y, test_size=0.3, random_state=0)
    
22. Standardizing the features:
    
23. 
    
24. In [6]:
    
25. from sklearn.preprocessing import StandardScaler
    
26. 
    
27. sc = StandardScaler()
    
28. sc.fit(X_train)
    
29. X_train_std = sc.transform(X_train)
    
30. X_test_std = sc.transform(X_test)

复制代码

1. Training a perceptron via scikit-learn
   
2. Redefining the plot_decision_region function from chapter 2:
   
3. In [7]:
   
4. from sklearn.linear_model import Perceptron
   
5. 
   
6. ppn = Perceptron(n_iter=40, eta0=0.1, random_state=0)
   
7. ppn.fit(X_train_std, y_train)
   
8. 
   
9. In [8]:
   
10. y_test.shape
    
11. 
    
12. In [9]:
    
13. y_pred = ppn.predict(X_test_std)
    
14. print('Misclassified samples: %d' % (y_test != y_pred).sum())
    
15. 
    
16. In [10]:
    
17. from sklearn.metrics import accuracy_score
    
18. 
    
19. print('Accuracy: %.2f' % accuracy_score(y_test, y_pred))
    
20. 
    
21. In [11]:
    
22. from matplotlib.colors import ListedColormap
    
23. import matplotlib.pyplot as plt
    
24. import warnings
    
25. 
    
26. 
    
27. def versiontuple(v):
    
28.     return tuple(map(int, (v.split("."))))
    
29. 
    
30. 
    
31. def plot_decision_regions(X, y, classifier, test_idx=None, resolution=0.02):
    
32. 
    
33.     # setup marker generator and color map
    
34.     markers = ('s', 'x', 'o', '^', 'v')
    
35.     colors = ('red', 'blue', 'lightgreen', 'gray', 'cyan')
    
36.     cmap = ListedColormap(colors[:len(np.unique(y))])
    
37. 
    
38.     # plot the decision surface
    
39.     x1_min, x1_max = X[:, 0].min() - 1, X[:, 0].max() + 1
    
40.     x2_min, x2_max = X[:, 1].min() - 1, X[:, 1].max() + 1
    
41.     xx1, xx2 = np.meshgrid(np.arange(x1_min, x1_max, resolution),
    
42.                            np.arange(x2_min, x2_max, resolution))
    
43.     Z = classifier.predict(np.array([xx1.ravel(), xx2.ravel()]).T)
    
44.     Z = Z.reshape(xx1.shape)
    
45.     plt.contourf(xx1, xx2, Z, alpha=0.4, cmap=cmap)
    
46.     plt.xlim(xx1.min(), xx1.max())
    
47.     plt.ylim(xx2.min(), xx2.max())
    
48. 
    
49.     for idx, cl in enumerate(np.unique(y)):
    
50.         plt.scatter(x=X[y == cl, 0],
    
51.                     y=X[y == cl, 1],
    
52.                     alpha=0.6,
    
53.                     c=cmap(idx),
    
54.                     edgecolor='black',
    
55.                     marker=markers[idx],
    
56.                     label=cl)
    
57. 
    
58.     # highlight test samples
    
59.     if test_idx:
    
60.         # plot all samples
    
61.         if not versiontuple(np.__version__) >= versiontuple('1.9.0'):
    
62.             X_test, y_test = X[list(test_idx), :], y[list(test_idx)]
    
63.             warnings.warn('Please update to NumPy 1.9.0 or newer')
    
64.         else:
    
65.             X_test, y_test = X[test_idx, :], y[test_idx]
    
66. 
    
67.         plt.scatter(X_test[:, 0],
    
68.                     X_test[:, 1],
    
69.                     c='',
    
70.                     alpha=1.0,
    
71.                     edgecolor='black',
    
72.                     linewidths=1,
    
73.                     marker='o',
    
74.                     s=55, label='test set')
    
75. Training a perceptron model using the standardized training data:
    
76. In [12]:
    
77. X_combined_std = np.vstack((X_train_std, X_test_std))
    
78. y_combined = np.hstack((y_train, y_test))
    
79. 
    
80. plot_decision_regions(X=X_combined_std, y=y_combined,
    
81.                       classifier=ppn, test_idx=range(105, 150))
    
82. plt.xlabel('petal length [standardized]')
    
83. plt.ylabel('petal width [standardized]')
    
84. plt.legend(loc='upper left')
    
85. 
    
86. plt.tight_layout()
    
87. # plt.savefig('./figures/iris_perceptron_scikit.png', dpi=300)
    
88. plt.show()

复制代码

1. Logistic regression intuition and conditional probabilities
   
2. In [13]:
   
3. import matplotlib.pyplot as plt
   
4. import numpy as np
   
5. 
   
6. 
   
7. def sigmoid(z):
   
8.     return 1.0 / (1.0 + np.exp(-z))
   
9. 
   
10. z = np.arange(-7, 7, 0.1)
    
11. phi_z = sigmoid(z)
    
12. 
    
13. plt.plot(z, phi_z)
    
14. plt.axvline(0.0, color='k')
    
15. plt.ylim(-0.1, 1.1)
    
16. plt.xlabel('z')
    
17. plt.ylabel('$\phi (z)$')
    
18. 
    
19. # y axis ticks and gridline
    
20. plt.yticks([0.0, 0.5, 1.0])
    
21. ax = plt.gca()
    
22. ax.yaxis.grid(True)
    
23. 
    
24. plt.tight_layout()
    
25. # plt.savefig('./figures/sigmoid.png', dpi=300)
    
26. plt.show()

复制代码

1. Learning the weights of the logistic cost function
   
2. In [15]:
   
3. def cost_1(z):
   
4.     return - np.log(sigmoid(z))
   
5. 
   
6. 
   
7. def cost_0(z):
   
8.     return - np.log(1 - sigmoid(z))
   
9. 
   
10. z = np.arange(-10, 10, 0.1)
    
11. phi_z = sigmoid(z)
    
12. 
    
13. c1 = [cost_1(x) for x in z]
    
14. plt.plot(phi_z, c1, label='J(w) if y=1')
    
15. 
    
16. c0 = [cost_0(x) for x in z]
    
17. plt.plot(phi_z, c0, linestyle='--', label='J(w) if y=0')
    
18. 
    
19. plt.ylim(0.0, 5.1)
    
20. plt.xlim([0, 1])
    
21. plt.xlabel('$\phi$(z)')
    
22. plt.ylabel('J(w)')
    
23. plt.legend(loc='best')
    
24. plt.tight_layout()
    
25. # plt.savefig('./figures/log_cost.png', dpi=300)
    
26. plt.show()

复制代码

1. Training a logistic regression model with scikit-learn
   
2. In [16]:
   
3. from sklearn.linear_model import LogisticRegression
   
4. 
   
5. lr = LogisticRegression(C=1000.0, random_state=0)
   
6. lr.fit(X_train_std, y_train)
   
7. 
   
8. plot_decision_regions(X_combined_std, y_combined,
   
9.                       classifier=lr, test_idx=range(105, 150))
   
10. plt.xlabel('petal length [standardized]')
    
11. plt.ylabel('petal width [standardized]')
    
12. plt.legend(loc='upper left')
    
13. plt.tight_layout()
    
14. # plt.savefig('./figures/logistic_regression.png', dpi=300)
    
15. plt.show()

复制代码

Thanks for sharing

kankan

谢谢分享