利用机器学习对信号特征分类

东方耀

利用机器学习对信号特征分类

1. 
   
2. import numpy as np
   
3. from sklearn.ensemble import RandomForestClassifier
   
4. from sklearn.metrics import classification_report
   
5. from sklearn.ensemble import GradientBoostingClassifier
   
6. # from siml.sk_utils import *
   
7. # from siml.signal_analysis_utils import *
   
8. 
   
9. # 利用机器学习对信号特征分类
   
10. # load提取的特征
    
11. 
    
12. train = np.load("train.npy")
    
13. X_train, Y_train = train[:, :-1], train[:, -1]
    
14. print(X_train.shape, Y_train.shape)
    
15. 
    
16. test = np.load("test.npy")
    
17. X_test, Y_test = test[:, :-1], test[:, -1]
    
18. print(X_test.shape, Y_test.shape)
    
19. 
    
20. # '''
    
21. clf = RandomForestClassifier(n_estimators=1000)
    
22. # clf = GradientBoostingClassifier()
    
23. 
    
24. clf.fit(X_train, Y_train)
    
25. # Accuracy on training set is : 1.0
    
26. # Accuracy on test set is : 0.8798778418730913
    
27. print("Accuracy on training set is : {}".format(clf.score(X_train, Y_train)))
    
28. print("Accuracy on test set is : {}".format(clf.score(X_test, Y_test)))
    
29. Y_test_pred = clf.predict(X_test)
    
30. print(classification_report(Y_test, Y_test_pred))
    
31. # '''
    
32. 
    
33. # See https://github.com/taspinar/siml
    
34. # dict_results = batch_classify(X_train, Y_train, X_test, Y_test)
    
35. # python3.8不支持clock了，替换成time.perf_counter()替换就可以了
    
36. # display_dict_models(dict_results)
    
37. 
    
38. 
    
39. # (7352, 270) (7352,)
    
40. # (2947, 270) (2947,)
    
41. # Accuracy on training set is : 1.0
    
42. # Accuracy on test set is : 0.8819138106549033
    
43. #               precision    recall  f1-score   support
    
44. #
    
45. #          1.0       0.93      0.98      0.96       496
    
46. #          2.0       0.93      0.95      0.94       471
    
47. #          3.0       0.96      0.88      0.92       420
    
48. #          4.0       0.80      0.75      0.77       491
    
49. #          5.0       0.78      0.93      0.85       532
    
50. #          6.0       0.93      0.81      0.87       537
    
51. #
    
52. #     accuracy                           0.88      2947
    
53. #    macro avg       0.89      0.88      0.88      2947
    
54. # weighted avg       0.89      0.88      0.88      2947
    
55. 
    

复制代码