【GitHub】Learning scikit-learn: Machine Learning in Python.

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

Learning scikit-learn: Machine Learning in PythonIpython sources for each chapter of the book

This repository holds all the ipython source and data for the "Learning scikit-learn: machine learning in Python" book, by Raúl Garreta and Guillermo Moncecchi (http://www.packtpub.com/learning-scikit-learn-machine-in-python/book). For the planned 2nd edition, we added Diego Garat as a new author.

• Chapter 1 - A Gentle Introduction to Machine Learning

• Chapter 1 (2nd ed.) - A Gentle Introduction to Machine Learning with Python and Scikit-learn - Extended version, including classification, clustering and regression!. Warning:Python 3

• [Chapter 2 - Supervised Learning - Image Recognition with Support Vector Machines] (http://nbviewer.ipython.org/github/gmonce/scikit-learn-book/blob/master/Chapter%202%20-%20Supervised%20Learning%20-%20Image%20Recognition%20with%20Support%20Vector%20Machines.ipynb)

• Chapter 2 - Supervised Learning - Regression

• [Chapter 2 - Supervised Learning - Text Classification with Naive Bayes] (http://nbviewer.ipython.org/github/gmonce/scikit-learn-book/blob/master/Chapter%202%20-%20Supervised%20Learning%20-%20Text%20Classification%20with%20Naive%20Bayes.ipynb)

• [Chapter 2 - Supervised Learning - Explaining Titanic Hypothesis with Decision Trees] (http://nbviewer.ipython.org/github/gmonce/scikit-learn-book/blob/master/Chapter%202%20-%20Supervised%20learning%20-%20Explaining%20Titanic%20Hypothesis%20with%20Decision%20Trees.ipynb)

• Chapter 3 - Unsupervised Learning - Clustering Handwritten Digits

• Chapter 3 - Unsupervised Learning - Principal Component Analysis

• Chapter 4 - Advanced Features - Feature Engineering and Selection

• Chapter 4 - Advanced Features - Model Selection

  
  

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

关键词：learning machine source

1. K-Means
   
2. from sklearn import cluster
   
3. clf = cluster.KMeans(init='k-means++', n_clusters=10, random_state=42)
   
4. clf.fit(X_train)
   
5. print clf.labels_.shape
   
6. print clf.labels_[1:10]
   
7. print_digits(images_train, clf.labels_, max_n=10)
   
8. # Predict clusters on testing data
   
9. y_pred = clf.predict(X_test)
   
10. 
    
11. def print_cluster(images, y_pred, cluster_number):
    
12.     images = images[y_pred==cluster_number]
    
13.     y_pred = y_pred[y_pred==cluster_number]
    
14.     print_digits(images, y_pred, max_n=10)
    
15. 
    
16. for i in range(10):
    
17.      print_cluster(images_test, y_pred, i)
    
18. from sklearn import metrics
    
19. print "Addjusted rand score:{:.2}".format(metrics.adjusted_rand_score(y_test, y_pred))
    
20. print "Homogeneity score:{:.2} ".format(metrics.homogeneity_score(y_test, y_pred))
    
21. print "Completeness score: {:.2} ".format(metrics.completeness_score(y_test, y_pred))
    
22. print "Confusion matrix"
    
23. print metrics.confusion_matrix(y_test, y_pred)
    
24. from sklearn import decomposition
    
25. # in this case the seeding of the centers is deterministic, hence we run the
    
26. # kmeans algorithm only once with n_init=1
    
27. pca = decomposition.PCA(n_components=2).fit(X_train)
    
28. reduced_X_train = pca.transform(X_train)
    
29. # Step size of the mesh. Decrease to increase the quality of the VQ.
    
30. h = .01     # point in the mesh [x_min, m_max]x[y_min, y_max].
    
31. 
    
32. # Plot the decision boundary. For that, we will asign a color to each
    
33. x_min, x_max = reduced_X_train[:, 0].min() + 1, reduced_X_train[:, 0].max() - 1
    
34. y_min, y_max = reduced_X_train[:, 1].min() + 1, reduced_X_train[:, 1].max() - 1
    
35. xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
    
36. 
    
37. kmeans = cluster.KMeans(init='k-means++', n_clusters=n_digits, n_init=10)
    
38. kmeans.fit(reduced_X_train)
    
39. Z = kmeans.predict(np.c_[xx.ravel(), yy.ravel()])
    
40. # Put the result into a color plot
    
41. Z = Z.reshape(xx.shape)
    
42. plt.figure(1)
    
43. plt.clf()
    
44. plt.imshow(Z, interpolation='nearest',
    
45.           extent=(xx.min(), xx.max(), yy.min(), yy.max()),
    
46.           cmap=plt.cm.Paired,
    
47.           aspect='auto', origin='lower')
    
48. #print reduced_X_train.shape
    
49. 
    
50. plt.plot(reduced_X_train[:, 0], reduced_X_train[:, 1], 'k.', markersize=2)
    
51. # Plot the centroids as a white X
    
52. centroids = kmeans.cluster_centers_
    
53. 
    
54. plt.scatter(centroids[:, 0], centroids[:, 1],
    
55.            marker='.', s=169, linewidths=3,
    
56.            color='w', zorder=10)
    
57. 
    
58. plt.title('K-means clustering on the digits dataset (PCA-reduced data)\n'
    
59.          'Centroids are marked with white dots')
    
60. plt.xlim(x_min, x_max)
    
61. plt.ylim(y_min, y_max)
    
62. plt.xticks(())
    
63. plt.yticks(())
    
64. plt.show()

复制代码

1. In [8]:
   
2. aff = cluster.AffinityPropagation()
   
3. aff.fit(X_train)
   
4. print aff.cluster_centers_indices_.shape
   
5. 
   
6. In [9]:
   
7. print_digits(images_train[aff.cluster_centers_indices_], y_train[aff.cluster_centers_indices_], max_n=aff.cluster_centers_indices_.shape[0])
   
8. 
   
9. In [10]#MeanShift
   
10. ms = cluster.MeanShift()
    
11. ms.fit(X_train)
    
12. print ms.cluster_centers_
    
13. 
    
14. In [11]:print ms.cluster_centers_.shape

复制代码

1. from sklearn import mixture
   
2. 
   
3. # Define a heldout dataset to estimate covariance type
   
4. X_train_heldout, X_test_heldout, y_train_heldout, y_test_heldout = train_test_split(
   
5.         X_train, y_train,test_size=0.25, random_state=42)
   
6. for covariance_type in ['spherical','tied','diag','full']:
   
7.     gm=mixture.GMM(n_components=n_digits, covariance_type=covariance_type, random_state=42, n_init=5)
   
8.     gm.fit(X_train_heldout)
   
9.     y_pred=gm.predict(X_test_heldout)
   
10.     print "Adjusted rand score for covariance={}:{:.2}".format(covariance_type, metrics.adjusted_rand_score(y_test_heldout, y_pred))
    
11. 
    
12. gm = mixture.GMM(n_components=n_digits, covariance_type='tied', random_state=42)
    
13. gm.fit(X_train)
    
14. 
    
15. # Print train clustering and confusion matrix
    
16. y_pred = gm.predict(X_test)
    
17. print "Addjusted rand score:{:.2}".format(metrics.adjusted_rand_score(y_test, y_pred))
    
18. print "Homogeneity score:{:.2} ".format(metrics.homogeneity_score(y_test, y_pred))
    
19. print "Completeness score: {:.2} ".format(metrics.completeness_score(y_test, y_pred))
    
20. for i in range(10):
    
21.      print_cluster(images_test, y_pred, i)
    
22. print "Confusion matrix"
    
23. print metrics.confusion_matrix(y_test, y_pred)
    
24. 
    
25. pl=plt
    
26. from sklearn import decomposition
    
27. # in this case the seeding of the centers is deterministic, hence we run the
    
28. # kmeans algorithm only once with n_init=1
    
29. pca = decomposition.PCA(n_components=2).fit(X_train)
    
30. reduced_X_train = pca.transform(X_train)
    
31. # Step size of the mesh. Decrease to increase the quality of the VQ.
    
32. h = .01     # point in the mesh [x_min, m_max]x[y_min, y_max].
    
33. 
    
34. # Plot the decision boundary. For that, we will asign a color to each
    
35. # Taken from
    
36. x_min, x_max = reduced_X_train[:, 0].min() + 1, reduced_X_train[:, 0].max() - 1
    
37. y_min, y_max = reduced_X_train[:, 1].min() + 1, reduced_X_train[:, 1].max() - 1
    
38. xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))
    
39. 
    
40. gm.fit(reduced_X_train)
    
41. #print np.c_[xx.ravel(),yy.ravel()]
    
42. Z = gm.predict(np.c_[xx.ravel(), yy.ravel()])
    
43. # Put the result into a color plot
    
44. Z = Z.reshape(xx.shape)
    
45. pl.figure(1)
    
46. pl.clf()
    
47. pl.imshow(Z, interpolation='nearest',
    
48.           extent=(xx.min(), xx.max(), yy.min(), yy.max()),
    
49.           cmap=pl.cm.Paired,
    
50.           aspect='auto', origin='lower')
    
51. #print reduced_X_train.shape
    
52. 
    
53. pl.plot(reduced_X_train[:, 0], reduced_X_train[:, 1], 'k.', markersize=2)
    
54. # Plot the centroids as a white X
    
55. centroids = gm.means_
    
56. 
    
57. pl.scatter(centroids[:, 0], centroids[:, 1],
    
58.            marker='.', s=169, linewidths=3,
    
59.            color='w', zorder=10)
    
60. 
    
61. pl.title('Mixture of gaussian models on the digits dataset (PCA-reduced data)\n'
    
62.          'Means are marked with white dots')
    
63. pl.xlim(x_min, x_max)
    
64. pl.ylim(y_min, y_max)
    
65. pl.xticks(())
    
66. pl.yticks(())
    
67. pl.show()

复制代码

谢谢楼主分享！