keras逻辑回归进行贷款风险预测

analyticsvidhya项目贷款风险预测

Dream Housing Finance公司处理所有房屋贷款。他们遍布所有城市，半城市和农村地区。客户在该公司确认客户的贷款资格后，首先申请房屋贷款。

问题
公司希望在填写在线申请表时，根据提供的客户详细信息自动进行贷款资格流程（实时）。这些详细信息包括性别，婚姻状况，教育程度，家属人数，收入，贷款金额，信用记录等。为了使这一过程自动化，他们遇到了一个问题，即要确定有资格获得贷款金额的客户群，以便他们可以专门针对这些客户。在这里，他们提供了部分数据集。

获取并分析原始数据

数据集：https://datahack.analyticsvidhya.com/contest/practice-problem-loan-prediction-iii/#ProblemStatement
注册并下载训练集，待预测集文件

观察数据有缺失值，df.dropna()删除有NaN的行只剩下2/3数据，看来不能删除，只好逐列填充了

预处理数据

先合并训练集和待预测集，统一重编码变量

# 取值转换为01，均值补全缺失值
df['Gender'] = df['Gender'].str.replace('Male', '0').replace('Female', '1').fillna('0.5') #pd.value_counts(df['Gender'])
df['Married'] = df['Married'].str.replace('No', '0').replace('Yes', '1').fillna('0.5') 
df['Dependents'] = df['Dependents'].fillna('0')
df['Self_Employed'] = df['Self_Employed'].str.replace('No', '0').replace('Yes', '1').fillna('0.11')
df['LoanAmount'] = df['LoanAmount'].fillna(df['LoanAmount'].mean())
df['Loan_Amount_Term'] = df['Loan_Amount_Term'].fillna(df['Loan_Amount_Term'].mean())
df['Credit_History'] = df['Credit_History'].fillna('0.83')
# 取值缩小到01之间
df['ApplicantIncome'] = df['ApplicantIncome']/df['ApplicantIncome'].max()
df['CoapplicantIncome'] = df['CoapplicantIncome']/df['CoapplicantIncome'].max()
df['LoanAmount'] = df['LoanAmount']/df['LoanAmount'].max()
df['Loan_Amount_Term'] = df['Loan_Amount_Term']/df['Loan_Amount_Term'].max()
# 定类定序列进行OneHot-Coding
x_OneHot_df = pd.get_dummies(data=df,columns=['Dependents', 'Education', 'Property_Area'])
# pd转np
ndarray = x_OneHot_df.values
#将数据分为变量和应变量
X = np.delete(ndarray, (0,9), axis=1)#删除编号列和结果列
Y = ndarray[:614,9] #614为训练集与待预测集分界

pre_X = X[614:]
X = X[:614]
print(X.shape, Y.shape, pre_X.shape)

创建模型并训练

model = Sequential([
    Dense(8, activation='relu', input_dim=17),
    Dense(8, activation='relu'),
    Dense(1, activation='sigmoid'),
])
model.compile(optimizer='adam',
              loss='binary_crossentropy',
              metrics=['accuracy'])
history = model.fit(x=X, y=Y, validation_split=0.3, epochs=30, batch_size=16)

调试后训练过程展示如下

acc = history.history['accuracy']
val_acc = history.history['val_accuracy']

loss=history.history['loss']
val_loss=history.history['val_loss']

epochs_range = range(30)

plt.figure(figsize=(8, 8))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, acc, label='Training Accuracy')
plt.plot(epochs_range, val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')

plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss')
plt.plot(epochs_range, val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()

预测并保存预测结果

pre_Y = model.predict(pre_X)
result = []
for i in pre_Y:
    if i < 0.5:
        result.append('拒绝贷款')
    else:
        result.append('通过贷款')
df.iloc[614:, -1] = result
pre_df = df.iloc[614:]
pd.value_counts(pre_df.Loan_Status)


pre_df.to_excel('pre_result.xlsx')
# 通过贷款    299
# 拒绝贷款     68