[已解决]有关信用卡欺诈的那个例子

Astray.R

附件那边上传大小必须2m以下，我就只取了csv前面几行数据

代码是这样写的：

1. #信用卡欺诈检测
   
2. import pandas as pd
   
3. import matplotlib.pyplot as plt
   
4. import numpy as np
   
5. #%matplotlib inline 魔法方法ipython中用
   
6. data = pd.read_csv(r"D:\储存\学习\唐宇迪python\01、python数据分析与机器学习实战\课程资料\唐宇迪-机器学习课程资料\机器学习算法配套案例实战\逻辑回归-信用卡欺诈检测\creditcard.csv")
   
7. count_classes = pd.value_counts(data['Class'], sort = True).sort_index()
   
8. count_classes.plot(kind = 'bar')                #建立柱状图
   
9. plt.title("Fraud class histogram")
   
10. plt.xlabel("Class")
    
11. plt.ylabel("Frequency")
    
12. 
    
13. from sklearn.preprocessing import StandardScaler
    
14. 
    
15. data['normAmount'] = StandardScaler().fit_transform(data['Amount'].values.reshape(-1, 1))
    
16. data = data.drop(['Time','Amount'],axis=1)
    
17. 
    
18. X = data.ix[:, data.columns != 'Class']
    
19. y = data.ix[:, data.columns == 'Class']
    
20. 
    
21. number_records_fraud = len(data[data.Class == 1])
    
22. fraud_indices = np.array(data[data.Class == 1].index)
    
23. 
    
24. 
    
25. normal_indices = data[data.Class == 0].index
    
26. 
    
27. random_normal_indices = np.random.choice(normal_indices, number_records_fraud, replace = False)
    
28. random_normal_indices = np.array(random_normal_indices)
    
29. 
    
30. under_sample_indices = np.concatenate([fraud_indices,random_normal_indices])
    
31. 
    
32. under_sample_data = data.iloc[under_sample_indices,:]
    
33. 
    
34. X_undersample = under_sample_data.ix[:, under_sample_data.columns != 'Class']
    
35. y_undersample = under_sample_data.ix[:, under_sample_data.columns == 'Class']
    
36. # Showing ratio
    
37. print("Percentage of normal transactions: ", len(under_sample_data[under_sample_data.Class == 0])/len(under_sample_data))
    
38. print("Percentage of fraud transactions: ", len(under_sample_data[under_sample_data.Class == 1])/len(under_sample_data))
    
39. print("Total number of transactions in resampled data: ", len(under_sample_data))
    
40. 
    
41. from sklearn.model_selection import train_test_split
    
42. 
    
43. 
    
44. X_train, X_test, y_train, y_test = train_test_split(X,y,test_size = 0.3, random_state = 0)
    
45. 
    
46. 
    
47. print("Number transactions train dataset: ", len(X_train))
    
48. print("Number transactions test dataset: ", len(X_test))
    
49. print("Total number of transactions: ", len(X_train)+len(X_test))
    
50. 
    
51. # 下采样
    
52. X_train_undersample, X_test_undersample, y_train_undersample, y_test_undersample = train_test_split(X_undersample,y_undersample,test_size = 0.3,random_state = 0)
    
53. print("")
    
54. print("Number transactions train dataset: ", len(X_train_undersample))
    
55. print("Number transactions test dataset: ", len(X_test_undersample))
    
56. print("Total number of transactions: ", len(X_train_undersample)+len(X_test_undersample))
    
57. 
    
58. #Recall = TP/(TP+FN)
    
59. from sklearn.linear_model import LogisticRegression                                
    
60. from sklearn.model_selection import KFold                                         
    
61. from sklearn.model_selection import cross_val_score                                
    
62. from sklearn.metrics import confusion_matrix,recall_score,classification_report   
    
63. 
    
64. #交叉验证
    
65. def printing_Kfold_scores(x_train_data,y_train_data):
    
66. 
    
67.     fold = KFold(5,shuffle=False)            
    
68. #    fold.get_n_splits(len(y_train_data))   
    
69.    
    
70.     # 定义不同正则化惩罚力度
    
71.     c_param_range = [0.01,0.1,1,10,100]
    
72. 
    
73.     #可视化显示
    
74.     results_table = pd.DataFrame(index = range(len(c_param_range),2), columns = ['C_parameter','Mean recall score'])
    
75.     results_table['C_parameter'] = c_param_range
    
76. 
    
77.     # the k-fold will give 2 lists: train_indices = indices[0], test_indices = indices[1]
    
78.     j = 0
    
79.     for c_param in c_param_range:
    
80.         print('-------------------------------------------')
    
81.         print('C parameter: ', c_param)
    
82.         print('-------------------------------------------')
    
83.         print('')
    
84. 
    
85.         recall_accs = []
    
86.         for iteration, indices in fold.split(X):   
    
87.             lr = LogisticRegression(C = c_param, penalty = 'l1')
    
88. 
    
89.             # Use the training data to fit the model. In this case, we use the portion of the fold to train the model
    
90.             # with indices[0]. We then predict on the portion assigned as the 'test cross validation' with indices[1]
    
91.             lr.fit(x_train_data.iloc[indices[0],:],y_train_data.iloc[indices[0],:].values.ravel())
    
92. 
    
93.             # Predict values using the test indices in the training data
    
94.             y_pred_undersample = lr.predict(x_train_data.iloc[indices[1],:].values)
    
95. 
    
96.             # Calculate the recall score and append it to a list for recall scores representing the current c_parameter
    
97.             recall_acc = recall_score(y_train_data.iloc[indices[1],:].values,y_pred_undersample)
    
98.             recall_accs.append(recall_acc)
    
99.             print('Iteration ', iteration,': recall score = ', recall_acc)
    
100. 
     
101.         # The mean value of those recall scores is the metric we want to save and get hold of.
     
102.         results_table.ix[j,'Mean recall score'] = np.mean(recall_accs)
     
103.         j += 1
     
104.         print('')
     
105.         print('Mean recall score ', np.mean(recall_accs))
     
106.         print('')
     
107. 
     
108.     best_c = results_table.loc[results_table['Mean recall score'].idxmax()]['C_parameter']
     
109.    
     
110.     # Finally, we can check which C parameter is the best amongst the chosen.
     
111.     print('*********************************************************************************')
     
112.     print('Best model to choose from cross validation is with C parameter = ', best_c)
     
113.     print('*********************************************************************************')
     
114.    
     
115.     return best_c
     
116. 
     
117. best_c = printing_Kfold_scores(X_train_undersample,y_train_undersample)

复制代码

===========================================================================================

结果就返回错误：
ValueError: Expected 2D array, got 1D array instead:
array=[-1.86375555  3.44264398 -4.46825973  2.80533626 -2.11841248 -2.33228489
-4.2612372   1.70168184 -1.43939588 -6.99990663  6.31620968 -8.670818
  0.31602399 -7.41771206 -0.43653747 -3.65280196 -6.29314532 -1.24324829
  0.36481048  0.360924    0.66792657 -0.51624236 -0.01221781  0.0706137
  0.05850447  0.30488284  0.41801247  0.20885828 -0.34923131].
Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample.

是这行有什么问题嘛？辛苦各位大神帮忙看看啦~
lr.fit(x_train_data.iloc[indices[0],:],y_train_data.iloc[indices[0],:].values.ravel())

最佳答案

月排行榜 / 总排行榜

塔利班

2019-12-11 16:23:58

维度不对，期待的是二维矩阵，而且都提示将其reshape下就行，至于哪个维度该是1，你自己跑的应该更明白

跳转到最佳答案楼层

塔利班

维度不对，期待的是二维矩阵，而且都提示将其reshape下就行，至于哪个维度该是1，你自己跑的应该更明白

Astray.R

塔利班 发表于 2019-12-11 16:36
维度不对，期待的是二维矩阵，而且都提示将其reshape下就行，至于哪个维度该是1，你自己跑的应该更明白

加了个冒号，维度应该没问题了，可是又提示
ValueError: Found array with 0 sample(s) (shape=(0, 29)) while a minimum of 1 is required.
这又是怎么回事呢？
我查了下如果加填个print(y_train_data.iloc[indices[0]:,:].values.ravel())
他只有第一次有数据，后面循环得到的全都是[]

塔利班

Astray.R 发表于 2019-12-11 21:35
加了个冒号，维度应该没问题了，可是又提示
ValueError: Found array with 0 sample(s) (shape=(0, 29)) ...

没时间跑你的数据，你把numpy数组搞明白就行，还有你的indices也搞明白

Astray.R

解决办法：
for iteration, indices in fold.split(X)改成
for iteration, indices in enumerate(fold.split(x_train_data), start=1)

小han

X = data.ix[:, data.columns != 'Class']
y = data.ix[:, data.columns == 'Class']

小han

X = data.ix[:, data.columns != 'Class']
y = data.ix[:, data.columns == 'Class']
这两段代码该怎么理解呢