5.4 logistic回归分析

logistic回归分析

logistic回归分析主要用来做分类（尤其是二分类问题），下面以客户是否会及时还信用卡欠款的数据进行预测

https://archive.ics.uci.edu/ml/datasets/default+of+credit+card+clients

信用卡是否违约数据集

from sklearn.linear_model import LogisticRegression

## 读取数据集
credit = pd.read_excel("D:\Desktop\python在机器学习中的应用\default of credit card clients.xls")
credit.head(5)

Y：1代表还款，0代表未还款，共计30000个样本，Y=1的样本约有6600个

## 检查是否有缺失值
credit.info()

## 6636个不违约，其余的违约
credit["Y"].sum()

## 数据集切分
trainx = ['X1', 'X2', 'X3', 'X4', 'X5', 'X6', 'X7', 'X8', 'X9', 'X10',
       'X11', 'X12', 'X13', 'X14', 'X15', 'X16', 'X17', 'X18', 'X19', 'X20',
       'X21', 'X22', 'X23']
Target = ["Y"]
##将训练集切分为训练集和验证集
traindata_x,valdata_x,traindata_y,valdata_y = train_test_split(credit[trainx], 
                                                               credit[Target],test_size = 0.25,random_state = 1)
traindata_x.sample(5)

clf_l1_LR = LogisticRegression(penalty = 'l1',solver='liblinear') 
#penalty='l1'，代表模型使用l1范数来约束自变量，和lasso回归中的惩罚函数一样，都有对自变量的控制作用
clf_l1_LR.fit(traindata_x,traindata_y)
pre_y = clf_l1_LR.predict(valdata_x)
print(metrics.classification_report(valdata_y,pre_y))

1. precision：预测为正的样本中有多少是真正的正样本
2. recall：召回率，样本中的正例有多少被预测正确了
3. f1-score：综合评价指标
4. support：预测为相应类别的数量

## plot ROC曲线
pre_y_p = clf_l1_LR.predict_proba(valdata_x)[:, 1]

fpr_LR, tpr_LR, _ = metrics.roc_curve(valdata_y, pre_y_p)
auc = metrics.auc(fpr_LR, tpr_LR)
plt.figure(1)
plt.plot([0, 1], [0, 1], 'k--')
plt.plot(fpr_LR, tpr_LR,"r",linewidth = 2)
plt.xlabel('False positive rate')
plt.ylabel('True positive rate')
plt.xlim(0, 1)
plt.ylim(0, 1)
plt.title('Logistic ROC curve')
plt.text(0.2,0.8,"auc = "+str(round(auc,4)))
plt.show()