Training Perception with scikit-learn 

 1 from sklearn import datasets
 2 import numpy as np
 3 
 4 iris = datasets.load_iris()
 5 
 6 X = iris.data[:, [2, 3]]
 7 y = iris.target
 8 print('Class labels:', np.unique(y))
 9 
10 from sklearn.model_selection import train_test_split
11 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.3, random_state = 1, stratify = y)
12 
13 print('Labels counts in y:', np.bincount(y))
14 print('Labels counts in y_train:', np.bincount(y_train))
15 print('Labels counts in y_test:', np.bincount(y_test))
16 
17 from sklearn.preprocessing import StandardScaler
18 sc = StandardScaler()
19 sc.fit(X_train)
20 X_train_std = sc.transform(X_train)
21 X_test_std = sc.transform(X_test)
22 
23 from sklearn.linear_model import Perceptron
24 
25 ppn = Perceptron(max_iter=40, eta0=0.1, random_state = 1)
26 ppn.fit(X_train_std, y_train)
27 
28 y_pred = ppn.predict(X_test_std)
29 print('Misclassified samples: %d' % (y_test != y_pred).sum())
30 
31 from sklearn.metrics import accuracy_score
32 print('Accuracy: %.2f' % accuracy_score(y_test, y_pred))
33 print('Accuracy: %.2f' % ppn.score(X_test_std, y_test))
34 
35 
36 import matplotlib.pyplot as plt
37 from matplotlib.colors import ListedColormap
38 
39 def plot_decision_regions(X, y, classifier, test_idx = None, resolution = 0.02):
40     #setup marker generator and color map
41     markers = ('s', 'x', 'o', '^', 'v')
42     colors = ('red', 'blue', 'lightgreen', 'gray', 'cyan')
43     cmap = ListedColormap(colors[:len(np.unique(y))])
44     
45     #plot the decision surface
46     x1_min, x1_max = X[:, 0].min() - 1, X[:, 0].max() + 1
47     x2_min, x2_max = X[:, 1].min() - 1, X[:, 1].max() + 1   
48     xx1, xx2 = np.meshgrid(np.arange(x1_min, x1_max, resolution),
49                           np.arange(x2_min, x2_max, resolution))
50     Z = classifier.predict(np.array([xx1.ravel(), xx2.ravel()]).T)
51     Z = Z.reshape(xx1.shape)
52     plt.contourf(xx1, xx2, Z, alpha=0.3, cmap = cmap)
53     plt.xlim(xx1.min(), xx2.max())
54     plt.ylim(xx2.min(), xx2.max())
55     
56     for idx, cl in enumerate(np.unique(y)):
57         plt.scatter(x=X[y == cl, 0], y=X[y == cl, 1], 
58                    alpha=0.8, c=colors[idx],
59                    marker = markers[idx], label=cl,
60                    edgecolor='black')
61     
62     #highlight test samples
63     if test_idx:
64         #plot all samples
65         X_test, y_test = X[test_idx, :], y[test_idx]
66         
67         plt.scatter(X_test[:, 0], X_test[:, 1], c='', 
68                    edgecolor = 'black', alpha=1.0, 
69                    linewidth=1, marker='o', s=100, 
70                    label='test set')
71         
72 X_combined_std = np.vstack((X_train_std, X_test_std))
73 y_combined = np.hstack((y_train, y_test))
74 plot_decision_regions(X=X_combined_std, 
75                      y=y_combined,
76                      classifier = ppn,
77                      test_idx = range(105, 150)
78                      )
79 plt.xlabel('petal length [standardized]')
80 plt.ylabel('petal width [standardized]')
81 plt.legend(loc='upper left')
82 plt.show()

 

posted @ 2019-06-08 22:09  WWBlog  阅读(271)  评论(0)    收藏  举报