import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

%matplotlib inline

data = pd.read_scv("creditcard.csv")
data.head()

data

help(pd.value_counts) # 計(jì)算非空值計(jì)數(shù)的直方圖

value_counts(values, sort=True, ascending=False, normalize=False, bins=None, dropna=True) Compute a histogram of the counts of non-null values.

count_classes = pd.value_counts(data['Class'], sort = True).sort_index()
# obj.sort_index() 按行索引進(jìn)行排序
count_classes.plot(kind = 'bar')
plt.title("Fraud class histogram")
plt.xlabel("Class")
plt.ylabel("Frequency")

交易正常與異常的數(shù)量

針對(duì)不平衡樣本處理方法：過(guò)采樣和欠采樣

X=data.loc[:, data.columns != 'Class'] # 將不是class列的數(shù)據(jù)賦給X
y = data.loc[:, data.columns == 'Class'] 
# 計(jì)算交易異常的個(gè)數(shù)
number_records_fraud = len(data[data.Class == 1])
# data.Class ==1 返回的是布爾值，data[True]是返回Class列為1的整行數(shù)據(jù)
fraud_indices = np.array(data[data.Class == 1 ].index)
# 將Class為1的數(shù)據(jù)的索引生成一個(gè)數(shù)組峭范。
normal_indices = data[data.Class == 0].index
# 將Class = 0 的數(shù)據(jù)的索引賦給normal_indices

欠采樣

random_normal_indices = np.random.choice(normal_indices, number_records_fraud, replace = False)
# 隨機(jī)選擇（從Class = 0 的索引里選擇，數(shù)量和class = 1 的一樣多，是否替換樣本為否）
random_normal_indices = np.array(random_normal_indices)

# Appending the 2 indices
under_sample_indices = np.concatenate([fraud_indices,random_normal_indices])
## 將兩個(gè)索引數(shù)組拼接起來(lái)
# Under sample dataset
under_sample_data = data.iloc[under_sample_indices,:]
# 通過(guò)索引將每一行數(shù)據(jù)賦給欠采樣數(shù)據(jù)集
X_undersample = under_sample_data.loc[:, under_sample_data.columns != 'Class']
y_undersample = under_sample_data.loc[:, under_sample_data.columns == 'Class']
# 將欠采樣數(shù)據(jù)集的X,y分開(kāi)

# Showing ratio 計(jì)算正常交易和異常交易的所占的比例
print("Percentage of normal transactions: ", len(under_sample_data[under_sample_data.Class == 0])/len(under_sample_data))
print("Percentage of fraud transactions: ", len(under_sample_data[under_sample_data.Class == 1])/len(under_sample_data))
print("Total number of transactions in resampled data: ", len(under_sample_data))

欠采樣樣本

loc函數(shù)完全基于標(biāo)簽位置的索引器厘惦，所謂標(biāo)簽败富，位置就是定義的行標(biāo)題

iloc函數(shù)完全基于行號(hào)的索引器歹河，所謂的行號(hào)就是第0，1赖临，2行

from sklearn.model_selection import train_test_split  
#從sklearn庫(kù)調(diào)特征選擇將樣本分成為訓(xùn)練集和測(cè)試集
X_train, X_test, y_train, y_test = train_test_split(X，y, test_size = 0.3, random_state = 0) 
# 測(cè)試集占總樣本的0.3灾锯， 計(jì)算測(cè)試的個(gè)數(shù)兢榨。
# whole dataset
print("Number transactions train dataset:", len(X_train))
print("Number transactions test dataset:", len(X_test))
print("Total number of transactions:", len(X_train) + len(X_test))
# undersampled dataset
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)
print("")
print("Number transactions train dataset:", len(X_train_undersample))
print("Number transactions test dataset:", len(X_train_undersample))
print("Total number of transactions:", len(X_train_undersample) + len(X_test_undersample))

切分訓(xùn)練集和測(cè)試集

幾個(gè)評(píng)價(jià)指標(biāo)

評(píng)價(jià)指標(biāo)

錯(cuò)誤率

準(zhǔn)確度

精確度

召回率

F1_score

ROC曲線

以FPR為橫坐標(biāo)，TPR為縱坐標(biāo)顺饮，那么ROC曲線就是改變各種閾值后得到的所有坐標(biāo)點(diǎn) (FPR,TPR) 的連線吵聪，畫出來(lái)如下。紅線是隨機(jī)亂猜情況下的ROC兼雄，曲線越靠左上角吟逝，分類器越佳。

ROC

簡(jiǎn)單說(shuō)：AUC值越大的分類器鹅巍，正確率越高千扶。

#Recall = TP/(TP+FN)
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import KFold
from sklearn.metrics import confusion_matrix, recall_score, classification_report
# 分類報(bào)告sklearn中的classification_report函數(shù)用于顯示主要分類指標(biāo)的文本報(bào)告．在報(bào)告中顯示每個(gè)類的精確度，召回率骆捧，F(xiàn)1值等信息澎羞。 
# 主要參數(shù): 
# y_true：1維數(shù)組，或標(biāo)簽指示器數(shù)組/稀疏矩陣敛苇，目標(biāo)值妆绞。 
# y_pred：1維數(shù)組顺呕，或標(biāo)簽指示器數(shù)組/稀疏矩陣，分類器返回的估計(jì)值括饶。 
# labels：array株茶，shape = [n_labels]，報(bào)表中包含的標(biāo)簽索引的可選列表图焰。 
# target_names：字符串列表启盛，與標(biāo)簽匹配的可選顯示名稱（相同順序）。 
# sample_weight：類似于shape = [n_samples]的數(shù)組技羔，可選項(xiàng)僵闯，樣本權(quán)重。 
# digits：int藤滥，輸出浮點(diǎn)值的位數(shù)．

def printing_Kfold_scores(x_train_data, y_train_data):
    fold = KFold(5, shuffle = False) #進(jìn)行k折交叉驗(yàn)證鳖粟，是否在將數(shù)據(jù)分成批之前進(jìn)行洗牌為否
# 不同的罰參數(shù)
c_param_range = [0.01, 0.1, 1, 10, 100]
 results_table = pd.DataFrame(index = range(len(c_param_range),2), columns = ['C_parameter','Mean recall score'])
    #          形成DataFrame結(jié)構(gòu)表  索引為五行兩列，列名為C_parameter , Mean recall score
    results_table['C_parameter'] = c_param_range  # 將罰參數(shù)賦給C_parameter這一列

    # the k-fold will give 2 lists: train_indices = indices[0], test_indices = indices[1]
    j = 0
    for c_param in c_param_range:
        print('-------------------------------------------')
        print('C parameter: ', c_param)
        print('-------------------------------------------')
        print('')

        recall_accs = []    # 召回率
        for iteration, indices in enumerate(fold.split(x_train_data)):

            # Call the logistic regression model with a certain C parameter
            lr = LogisticRegression(C = c_param, penalty = 'l1')  # 調(diào)用邏輯回歸拙绊，L1型罰函數(shù)

            # 利用訓(xùn)練數(shù)據(jù)擬合模型向图。在這種情況下，我們使用折疊部分來(lái)訓(xùn)練模型
            # with indices[0]. We then predict on the portion assigned as the 'test cross validation' with indices[1]
            lr.fit(x_train_data.iloc[indices[0],:],y_train_data.iloc[indices[0],:].values.ravel())
            # rvavel(散開(kāi)标沪，解開(kāi))寿桨，flatten（變平）半火。兩者的區(qū)別在于返回拷貝（copy）還是返回視圖（view）微姊，
            #numpy.flatten()返回一份拷貝呕屎，對(duì)拷貝所做的修改不會(huì)影響（reflects）原始矩陣，
            # 而numpy.ravel()返回的是視圖趴梢，會(huì)影響（reflects）原始矩陣漠畜。
           
            # 使用訓(xùn)練數(shù)據(jù)中的測(cè)試指標(biāo)預(yù)測(cè)值
            y_pred_undersample = lr.predict(x_train_data.iloc[indices[1],:].values)

            # 計(jì)算召回率，并將其追加到表示當(dāng)前c_parameter的召回率列表中
            recall_acc = recall_score(y_train_data.iloc[indices[1],:].values,y_pred_undersample)
            recall_accs.append(recall_acc)
            print('Iteration ', iteration,': recall score = ', recall_acc)

        # The mean value of those recall scores is the metric we want to save and get hold of.
        results_table.loc[j,'Mean recall score'] = np.mean(recall_accs)
        j += 1
        print('')
        print('Mean recall score ', np.mean(recall_accs))
        print('')

    best_c=results_table.loc[results_table['Mean recall score'].astype('float64').idxmax()]['C_parameter']
     # 將召回率最大的罰參數(shù)輸出                                          
    # 最后坞靶，我們可以檢查選擇的C參數(shù)中哪個(gè)是最好的憔狞。
    print('*********************************************************************************')
    print('Best model to choose from cross validation is with C parameter = ', best_c)
    print('*********************************************************************************')
    
    return best_c

enumerate() 函數(shù)用于將一個(gè)可遍歷的數(shù)據(jù)對(duì)象(如列表、元組或字符串)組合為一個(gè)索引序列彰阴，同時(shí)列出數(shù)據(jù)和數(shù)據(jù)下標(biāo)瘾敢，

一般用在 for 循環(huán)當(dāng)中。

enumerate

import warnings
warnings.filterwarnings("ignore")  # 忽略警告
best_c = printing_Kfold_scores(X_train_undersample,y_train_undersample) # 將欠采樣數(shù)據(jù)集傳入進(jìn)去

不同罰參數(shù)

不同罰參數(shù)

不同罰參數(shù)

def plot_confusion_matrix(cm, classes, title = "Confusion matrix", camp = plt.cm.Blues):     # 混淆矩陣
plt.imshow(cm,interpolation = 'nearest', cmap = cmap)
plt.title(title)
plt.colorbar()
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, rotation = 0)
plt.yticks(tick_marks, classes)

thresh = cm.max()/2
for i, j in itertools.product(range(cm.shape[0]),range(cm.shape[1])):
    plt.text(j, i, cm[i, j], horizontalignment = "center", color = "white" if cm[i, j] > thresh else "black")

plt.tight_layout()
plt.ylabel('True label')
plt.xlabel("Predicted label")

import itertools
lr = LogisticRegression(C = best_c, penalty = 'l1')
lr.fit(X_train_undersample,y_train_undersample.values.ravel())
y_pred_undersample = lr.predict(X_test_undersample.values)

# 計(jì)算混淆矩陣
cnf_matrix = confusion_matrix(y_test_undersample,y_pred_undersample)
np.set_printoptions(precision=2)
print("Recall metric in the testing dataset: ", cnf_matrix[1,1]/(cnf_matrix[1,0]+cnf_matrix[1,1]))

# 繪制非標(biāo)準(zhǔn)化混淆矩陣
class_names = [0,1]
plt.figure()
plot_confusion_matrix(cnf_matrix
                      , classes=class_names
                      , title='Confusion matrix')
plt.show()

Confusion matrix

best_c = printing_Kfold_scores(X_train,y_train)

C parameter: 0.01

Iteration 0 : recall score = 0.4925373134328358
Iteration 1 : recall score = 0.6027397260273972
Iteration 2 : recall score = 0.6833333333333333
Iteration 3 : recall score = 0.5692307692307692
Iteration 4 : recall score = 0.45

Mean recall score 0.5595682284048672

C parameter: 0.1

Iteration 0 : recall score = 0.5671641791044776
Iteration 1 : recall score = 0.6164383561643836
Iteration 2 : recall score = 0.6833333333333333
Iteration 3 : recall score = 0.5846153846153846
Iteration 4 : recall score = 0.525

Mean recall score 0.5953102506435158

C parameter: 1

Iteration 0 : recall score = 0.5522388059701493
Iteration 1 : recall score = 0.6164383561643836
Iteration 2 : recall score = 0.7166666666666667
Iteration 3 : recall score = 0.6153846153846154
Iteration 4 : recall score = 0.5625

Mean recall score 0.612645688837163

C parameter: 10

Iteration 0 : recall score = 0.5522388059701493
Iteration 1 : recall score = 0.6164383561643836
Iteration 2 : recall score = 0.7333333333333333
Iteration 3 : recall score = 0.6153846153846154
Iteration 4 : recall score = 0.575

Mean recall score 0.6184790221704963

C parameter: 100

Iteration 0 : recall score = 0.5522388059701493
Iteration 1 : recall score = 0.6164383561643836
Iteration 2 : recall score = 0.7333333333333333
Iteration 3 : recall score = 0.6153846153846154
Iteration 4 : recall score = 0.575

Mean recall score 0.6184790221704963

Best model to choose from cross validation is with C parameter = 10.0

lr = LogisticRegression(C = best_c, penalty = 'l1')
lr.fit(X_train,y_train.values.ravel())
y_pred_undersample = lr.predict(X_test.values)

# Compute confusion matrix
cnf_matrix = confusion_matrix(y_test,y_pred_undersample)
np.set_printoptions(precision=2)

print("Recall metric in the testing dataset: ", cnf_matrix[1,1]/(cnf_matrix[1,0]+cnf_matrix[1,1]))

# Plot non-normalized confusion matrix
class_names = [0,1]
plt.figure()
plot_confusion_matrix(cnf_matrix
                      , classes=class_names
                      , title='Confusion matrix')
plt.show()

Recall metric in the testing dataset: 0.6190476190476191

image.png

lr = LogisticRegression(C = 0.01, penalty = 'l1')
lr.fit(X_train_undersample,y_train_undersample.values.ravel())
y_pred_undersample_proba = lr.predict_proba(X_test_undersample.values)

thresholds = [0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9]

plt.figure(figsize=(10,10))

j = 1
for i in thresholds:
    y_test_predictions_high_recall = y_pred_undersample_proba[:,1] > i
    
    plt.subplot(3,3,j)
    j += 1
    
    # Compute confusion matrix
    cnf_matrix = confusion_matrix(y_test_undersample,y_test_predictions_high_recall)
    np.set_printoptions(precision=2)

    print("Recall metric in the testing dataset: ", cnf_matrix[1,1]/(cnf_matrix[1,0]+cnf_matrix[1,1]))

    # Plot non-normalized confusion matrix
    class_names = [0,1]
    plot_confusion_matrix(cnf_matrix
                          , classes=class_names
                          , title='Threshold >= %s'%i) 

Recall metric in the testing dataset: 1.0
Recall metric in the testing dataset: 1.0
Recall metric in the testing dataset: 1.0
Recall metric in the testing dataset: 0.9727891156462585
Recall metric in the testing dataset: 0.9387755102040817
Recall metric in the testing dataset: 0.8979591836734694
Recall metric in the testing dataset: 0.8367346938775511
Recall metric in the testing dataset: 0.782312925170068
Recall metric in the testing dataset: 0.5986394557823129

untitled.png