进行数据展示
第一步我们先导入数据,查看一下要区分的类别得类别数,以及每个类别得样本数,并进行展示。信用贷款得CLASS只有两类,0和1,两者拥有的数据量差异极大,面临数据不平衡问题。
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import os
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression#引入逻辑回归分类器
from sklearn.model_selection import KFold, cross_val_score#引入K折交差验证
from sklearn.metrics import confusion_matrix,recall_score,classification_report #引入混淆矩阵,召回率,以及分类报告
import itertools
from imblearn.over_sampling import SMOTE
if __name__=="__main__":
path="D:\Python base\Test\逻辑回归-信用卡欺诈检测"+os.sep +"creditcard.csv"
data= pd.read_csv(path)
if __name__=="__main__":
path="D:\Python base\Test\逻辑回归-信用卡欺诈检测"+os.sep +"creditcard.csv"
data= pd.read_csv(path)
#1.1进行数据展示
class_number=data.loc[:,'Classnew'].value_counts()#统计某一列各个数值得出现次数
count_classes = pd.value_counts(data['Classnew'], sort = True).sort_index()#统计某一列各个数值得出现次数
class_number.plot(kind = 'bar')#kind可以是’line’, ‘bar’, ‘barh’, ‘kde’
plt.title("Fraud class histogram")
plt.xlabel("Class")
plt.ylabel("Frequency")
plt.show()
进行归一化和下采样
#对某列数据进行归一化
data['normAmount'] = StandardScaler().fit_transform(data['Amount'].values.reshape(-1, 1))#对某列数据进行归一化
data = data.drop(['Time','Amount'],axis=1)#删除没用的两列
X=data.loc[:,data.columns != "Classnew"]
y=data.loc[:,data.columns == "Classnew"]
#1.2进行下采样
class1=data[data.Classnew == 1] #利用条件充当Index挑选class=1的数据
print("class=1的数据为:",class1)
#class0=data[data["Classnew"] == 0] #利用
class1_number=len(class1)#统计class为1数据有多少
class1_index=np.array(data[data.Classnew == 1].index)
class0_index=np.array(data[data.Classnew == 0].index)
class0_indices = np.random.choice(class0_index, class1_number, replace = False)#随机选取class1_number个为零的数值索引
class0_indices = np.array(class0_indices)
under_sample_indices = np.concatenate([class1_index,class0_indices])#下采样的所有数据的Index
under_sample_data = data.iloc[under_sample_indices,:]#获取的下采样数据
X_undersample = under_sample_data.ix[:, under_sample_data.columns != 'Classnew']
y_undersample = under_sample_data.ix[:, under_sample_data.columns == 'Classnew']
print("Percentage of normal transactions: ", len(under_sample_data[under_sample_data.Classnew == 0])/len(under_sample_data))
print("Percentage of fraud transactions: ", len(under_sample_data[under_sample_data.Classnew == 1])/len(under_sample_data))
print("Total number of transactions in resampled data: ", len(under_sample_data))
原始数据和下采样数据(训练集和测试集的划分)
#训练集和验证集划分
X_train, X_test, y_train, y_test = train_test_split(X,y,test_size = 0.3, random_state = 0)
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)
K折交叉验证(对划分的下采样训练集数据)
假如train分为10分,前9分用于训练,后一份用于测试
#K折交叉验证
def printing_Kfold_scores(x_train_data,y_train_data):
#k折交叉验证
fold = KFold(n_splits=5,shuffle=False)
#不同的C参数
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'])
results_table['C_parameter'] = c_param_range
#k折操作将会给出两个列表: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 = []
#enumerate() 函数用于将一个可遍历的数据对象(如列表、元组或字符串)组合为一个索引序列,同时列出数据和数据下标,一般用在 for 循环当中。
for iteration,indices in enumerate(fold.split(x_train_data),start=1):
#把c_param_range代入到逻辑回归模型中,并使用了l1正则化,C代表正则化系数
#solver优化算法选择参数,{‘newton-cg’, ‘lbfgs’, ‘liblinear’, ‘sag’}, default: ‘liblinear’
lr = LogisticRegression(C = c_param,penalty = 'l1',solver='liblinear')
#使用indices[0]的数据进行拟合曲线,使用indices[1]的数据进行误差测试
lr.fit(x_train_data.iloc[indices[0],:],y_train_data.iloc[indices[0],:].values.ravel())#模型训练
#在indices[1]数据上预测值
y_pred_undersample = lr.predict(x_train_data.iloc[indices[1],:].values)
#根据不同的c_parameter计算召回率
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)
#求出我们想要的召回平均值
results_table.loc[j,'Mean recall score'] = np.mean(recall_accs)#计算召回率的平均值
j += 1
print('')
print('Mean recall score ', np.mean(recall_accs))
print('')
print("results_table is:",results_table) #输出每个参数c所对应的 K折交叉验证的平均召回率
best_c = results_table.loc[results_table['Mean recall score'].values.argmax()]['C_parameter']#返回最大平均召回率所对应的参数C
#最后选择最好的 C parameter
print('*********************************************************************************')
print('Best model to choose from cross validation is with C parameter = ', best_c)
print('*********************************************************************************')
return best_c
对于这个代码:
from sklearn.cross_validation import KFold
fold = KFold(50,5,shuffle=False)
for iteration, indices in enumerate(fold,start=1):
print(iteration, indices)
利用K折交叉验证返回的参数C进行模型的训练和预测
best_c = printing_Kfold_scores(X_train_undersample,y_train_undersample)
lr = LogisticRegression(C = best_c, penalty = 'l1',solver='liblinear')#创建模型
lr.fit(X_train_undersample,y_train_undersample.values.ravel())#模型预训练
y_pred_undersample = lr.predict(X_test_undersample.values)#模型预测
混淆矩阵(混淆矩阵的y轴代表真实值,X轴代表预测值)
TP代表真积极,T代表真,P代表正类;FP代表假积极,F代表假,P代表判为正类;FN代表假消极,F代表假,N代表负类;TN代表真负类,T代表判断正确,判为父类。召回率(recall)=TP/(TP+FN)
#混淆矩阵绘制
#cm为混淆矩阵,classes为混淆矩阵的类标
def plot_confusion_matrix(cm, classes,
title='Confusion matrix',
cmap=plt.cm.Blues):
"""
This function prints and plots the confusion matrix.
"""
plt.imshow(cm, interpolation='nearest', cmap=cmap)#interpolation插值方式,#cmap表示绘图时的样式,cm是混淆矩阵
plt.title(title)
plt.colorbar()#给图配渐变色时,常常需要在图旁边把colorbar显示出来
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, rotation=0)
plt.yticks(tick_marks, classes)
thresh = cm.max() / 2. #阈值
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, cm[i, j],
horizontalalignment="center",
color="white" if cm[i, j] > thresh else "black")
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
# Compute confusion matrix
cnf_matrix = confusion_matrix(y_test_undersample,y_pred_undersample)#计算混淆矩阵
np.set_printoptions(precision=2)#控制输出的小数点个数是2
print("混淆矩阵为:",cnf_matrix)
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()
查看不同阈值的混淆矩阵效果
#循环查看不通同阈值效果
lr = LogisticRegression(C = 0.01, penalty = 'l1',solver='liblinear')
lr.fit(X_train_undersample,y_train_undersample.values.ravel())
y_pred_undersample_proba = lr.predict_proba(X_test_undersample.values)#预测y值的概率
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)
标题
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import os
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression#引入逻辑回归分类器
from sklearn.model_selection import KFold, cross_val_score#引入K折交差验证
from sklearn.metrics import confusion_matrix,recall_score,classification_report #引入混淆矩阵,召回率,以及分类报告
import itertools
from imblearn.over_sampling import SMOTE
#过采样
features_train, features_test, labels_train, labels_test = train_test_split(X,
y,
test_size=0.2,
random_state=0)
oversampler=SMOTE(random_state=0)
os_features,os_labels=oversampler.fit_sample(features_train,labels_train)#对训练特征和标签数据过采样
print(len(os_labels[os_labels==1]))
print(len(os_labels[os_labels==0])
os_features = pd.DataFrame(os_features)
os_labels = pd.DataFrame(os_labels)
best_c = printing_Kfold_scores(os_features,os_labels)
整个代码
# -*- coding: utf-8 -*-
"""
Created on Fri Feb 28 07:07:23 2020
@author: User
"""
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import os
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression#引入逻辑回归分类器
from sklearn.model_selection import KFold, cross_val_score#引入K折交差验证
from sklearn.metrics import confusion_matrix,recall_score,classification_report #引入混淆矩阵,召回率,以及分类报告
import itertools
from imblearn.over_sampling import SMOTE
#K折交叉验证
def printing_Kfold_scores(x_train_data,y_train_data):
#k折交叉验证
fold = KFold(n_splits=5,shuffle=False)
#不同的C参数
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'])
results_table['C_parameter'] = c_param_range
#k折操作将会给出两个列表: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 = []
#enumerate() 函数用于将一个可遍历的数据对象(如列表、元组或字符串)组合为一个索引序列,同时列出数据和数据下标,一般用在 for 循环当中。
for iteration,indices in enumerate(fold.split(x_train_data),start=1):
#把c_param_range代入到逻辑回归模型中,并使用了l1正则化,C代表正则化系数
#solver优化算法选择参数,{‘newton-cg’, ‘lbfgs’, ‘liblinear’, ‘sag’}, default: ‘liblinear’
lr = LogisticRegression(C = c_param,penalty = 'l1',solver='liblinear')
#使用indices[0]的数据进行拟合曲线,使用indices[1]的数据进行误差测试
lr.fit(x_train_data.iloc[indices[0],:],y_train_data.iloc[indices[0],:].values.ravel())#模型训练
#在indices[1]数据上预测值
y_pred_undersample = lr.predict(x_train_data.iloc[indices[1],:].values)
#根据不同的c_parameter计算召回率
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)
#求出我们想要的召回平均值
results_table.loc[j,'Mean recall score'] = np.mean(recall_accs)#计算召回率的平均值
j += 1
print('')
print('Mean recall score ', np.mean(recall_accs))
print('')
print("results_table is:",results_table) #输出每个参数c所对应的 K折交叉验证的平均召回率
best_c = results_table.loc[results_table['Mean recall score'].values.argmax()]['C_parameter']#返回最大平均召回率所对应的参数C
#最后选择最好的 C parameter
print('*********************************************************************************')
print('Best model to choose from cross validation is with C parameter = ', best_c)
print('*********************************************************************************')
return best_c
#混淆矩阵绘制
#cm为混淆矩阵,classes为混淆矩阵的类标
def plot_confusion_matrix(cm, classes,
title='Confusion matrix',
cmap=plt.cm.Blues):
"""
This function prints and plots the confusion matrix.
"""
plt.imshow(cm, interpolation='nearest', cmap=cmap)#interpolation插值方式,#cmap表示绘图时的样式,cm是混淆矩阵
plt.title(title)
plt.colorbar()#给图配渐变色时,常常需要在图旁边把colorbar显示出来
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, rotation=0)
plt.yticks(tick_marks, classes)
thresh = cm.max() / 2. #阈值
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, cm[i, j],
horizontalalignment="center",
color="white" if cm[i, j] > thresh else "black")
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
if __name__=="__main__":
path="D:\Python base\Test\逻辑回归-信用卡欺诈检测"+os.sep +"creditcard.csv"
data= pd.read_csv(path)
#1.1进行数据展示
class_number=data.loc[:,'Classnew'].value_counts()#统计某一列各个数值得出现次数
count_classes = pd.value_counts(data['Classnew'], sort = True).sort_index()#统计某一列各个数值得出现次数
class_number.plot(kind = 'bar')#kind可以是’line’, ‘bar’, ‘barh’, ‘kde’
plt.title("Fraud class histogram")
plt.xlabel("Class")
plt.ylabel("Frequency")
plt.show()
#对某列数据进行归一化
data['normAmount'] = StandardScaler().fit_transform(data['Amount'].values.reshape(-1, 1))#对某列数据进行归一化
data = data.drop(['Time','Amount'],axis=1)#删除没用的两列
X=data.loc[:,data.columns != "Classnew"]
y=data.loc[:,data.columns == "Classnew"]
#1.2进行下采样
class1=data[data.Classnew == 1] #利用条件充当Index挑选class=1的数据
print("class=1的数据为:",class1)
#class0=data[data["Classnew"] == 0] #利用
class1_number=len(class1)#统计class为1数据有多少
class1_index=np.array(data[data.Classnew == 1].index)
class0_index=np.array(data[data.Classnew == 0].index)
class0_indices = np.random.choice(class0_index, class1_number, replace = False)#随机选取class1_number个为零的数值索引
class0_indices = np.array(class0_indices)
under_sample_indices = np.concatenate([class1_index,class0_indices])#下采样的所有数据的Index
under_sample_data = data.iloc[under_sample_indices,:]#获取的下采样数据
X_undersample = under_sample_data.ix[:, under_sample_data.columns != 'Classnew']
y_undersample = under_sample_data.ix[:, under_sample_data.columns == 'Classnew']
print("Percentage of normal transactions: ", len(under_sample_data[under_sample_data.Classnew == 0])/len(under_sample_data))
print("Percentage of fraud transactions: ", len(under_sample_data[under_sample_data.Classnew == 1])/len(under_sample_data))
print("Total number of transactions in resampled data: ", len(under_sample_data))
#
#训练集和验证集划分
X_train, X_test, y_train, y_test = train_test_split(X,y,test_size = 0.3, random_state = 0)
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)
features_train, features_test, labels_train, labels_test = train_test_split(X,
y,
test_size=0.2,
random_state=0)
oversampler=SMOTE(random_state=0)
os_features,os_labels=oversampler.fit_sample(features_train,labels_train)#对训练特征和标签数据过采样
print(len(os_labels[os_labels==1]))
print(len(os_labels[os_labels==0])
os_features = pd.DataFrame(os_features)
os_labels = pd.DataFrame(os_labels)
best_c = printing_Kfold_scores(os_features,os_labels)
best_c = printing_Kfold_scores(X_train_undersample,y_train_undersample)
lr = LogisticRegression(C = best_c, penalty = 'l1',solver='liblinear')#创建模型
lr.fit(X_train_undersample,y_train_undersample.values.ravel())#模型预训练
y_pred_undersample = lr.predict(X_test_undersample.values)#模型预测
# Compute confusion matrix
cnf_matrix = confusion_matrix(y_test_undersample,y_pred_undersample)#计算混淆矩阵
np.set_printoptions(precision=2)#控制输出的小数点个数是2
print("混淆矩阵为:",cnf_matrix)
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()
#循环查看不通同阈值效果
lr = LogisticRegression(C = 0.01, penalty = 'l1',solver='liblinear')
lr.fit(X_train_undersample,y_train_undersample.values.ravel())
y_pred_undersample_proba = lr.predict_proba(X_test_undersample.values)#预测y值的概率
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)
#