基于机器学习逻辑回归的信用评分算法

概述

Banks play a crucial role in market economies. They decide who can get finance and on what terms and can make or break investment decisions. For markets and society to function, individuals and companies need access to credit.

Credit scoring algorithms, which make a guess at the probability of default, are the method banks use to determine whether or not a loan should be granted. This competition requires participants to improve on the state of the art in credit scoring, by predicting the probability that somebody will experience financial distress in the next two years. Dataset

Attribute Information:

Variable Name	Description	Type
SeriousDlqin2yrs	Person experienced 90 days past due delinquency or worse	Y/N
RevolvingUtilizationOfUnsecuredLines	Total balance on credit divided by the sum of credit limits	percentage
age	Age of borrower in years	integer
NumberOfTime30-59DaysPastDueNotWorse	Number of times borrower has been 30-59 days past due	integer
DebtRatio	Monthly debt payments	percentage
MonthlyIncome	Monthly income	real
NumberOfOpenCreditLinesAndLoans	Number of Open loans	integer
NumberOfTimes90DaysLate	Number of times borrower has been 90 days or more past due.	integer
NumberRealEstateLoansOrLines	Number of mortgage and real estate loans	integer
NumberOfTime60-89DaysPastDueNotWorse	Number of times borrower has been 60-89 days past due	integer
NumberOfDependents	Number of dependents in family	integer

Read the data into Pandas

# 导入 pandas 库，并设置 pandas 的显示选项，使其能显示最多 500 列
# 使用 zipfile 库打开名为 'KaggleCredit2.csv.zip' 的 zip 文件，并从中读取 'KaggleCredit2.csv' 文件
# 显示 DataFrame 的头几行
import pandas as pd 
pd.set_option('display.max_columns', 500)
import zipfile
with zipfile.ZipFile('KaggleCredit2.csv.zip', 'r') as z:   ##读取zip里的文件
    f = z.open('KaggleCredit2.csv')
    data = pd.read_csv(f, index_col=0)
data.head()

# 显示数据的形状，即行数和列数
data.shape

(112915, 11)

Drop na

# 计算每一列中空值的数量
data.isnull().sum(axis=0)

SeriousDlqin2yrs                           0
RevolvingUtilizationOfUnsecuredLines       0
age                                     4267
NumberOfTime30-59DaysPastDueNotWorse       0
DebtRatio                                  0
MonthlyIncome                              0
NumberOfOpenCreditLinesAndLoans            0
NumberOfTimes90DaysLate                    0
NumberRealEstateLoansOrLines               0
NumberOfTime60-89DaysPastDueNotWorse       0
NumberOfDependents                      4267
dtype: int64

# 删除所有包含空值的行，然后再次显示数据的形状，以确认行数已经减少
data.dropna(inplace=True)   ##去掉为空的数据
data.shape

(108648, 11)

Create X and y

# 目标变量 'SeriousDlqin2yrs' 被从数据中分离出来，存储在 y 中
# 其余的特征存储在 X 中
y = data['SeriousDlqin2yrs']
X = data.drop('SeriousDlqin2yrs', axis=1)

# 计算目标变量的平均值
y.mean() ##求取均值

0.06742876076872101

练习1

把数据切分成训练集和测试集

# 数据被分成训练集和测试集，测试集的大小是总数据的 20%
from sklearn import model_selection
x_tran,x_test,y_tran,y_test=model_selection.train_test_split(X,y,test_size=0.2)
print(x_test.shape)

(21730, 10)

练习2

使用logistic regression/决策树/SVM/KNN…等sklearn分类算法进行分类，尝试查sklearn API了解模型参数含义，调整不同的参数。

# 创建 LogisticRegression 模型，并用训练数据对其进行拟合
# 计算模型在训练数据上的得分
from sklearn.linear_model import LogisticRegression
## https://blog.csdn.net/sun_shengyun/article/details/53811483
lr=LogisticRegression(multi_class='ovr',solver='sag',class_weight='balanced')
# lr=LogisticRegression(multi_class='ovr',solver='newton-cg',class_weight='balanced') #降低准确率 提升召回率
lr.fit(x_tran,y_tran)
score=lr.score(x_tran,y_tran)
print(score) ##最好的分数是1

0.9325801329989185


d:\Python\lib\site-packages\sklearn\linear_model\_sag.py:350: ConvergenceWarning: The max_iter was reached which means the coef_ did not converge
  warnings.warn(

练习3

在测试集上进行预测，计算准确度

# 计算模型在训练集和测试集上的准确率
from sklearn.metrics import accuracy_score
## https://blog.csdn.net/qq_16095417/article/details/79590455
train_score=accuracy_score(y_tran,lr.predict(x_tran))
test_score=lr.score(x_test,y_test)
print('训练集准确率：',train_score)
print('测试集准确率：',test_score)

训练集准确率： 0.9325801329989185
测试集准确率： 0.9324436263230557

练习4

查看sklearn的官方说明，了解分类问题的评估标准，并对此例进行评估。

# 计算模型在训练集和测试集上的召回率
##召回率
from sklearn.metrics import recall_score
train_recall=recall_score(y_tran,lr.predict(x_tran),average='macro')
test_recall=recall_score(y_test,lr.predict(x_test),average='macro')
print('训练集召回率：',train_recall)
print('测试集召回率：',test_recall)

训练集召回率： 0.5
测试集召回率： 0.4999506514015002

练习5

银行通常会有更严格的要求，因为fraud带来的后果通常比较严重，一般我们会调整模型的标准。

比如在logistic regression当中，一般我们的概率判定边界为0.5，但是我们可以把阈值设定低一些，来提高模型的“敏感度”，试试看把阈值设定为0.3，再看看这时的评估指标(主要是准确率和召回率)。

tips:sklearn的很多分类模型，predict_prob可以拿到预估的概率，可以根据它和设定的阈值大小去判断最终结果(分类类别)

# 计算模型对测试集中每个样本的预测概率
# 设置阈值（0.02），并将大于该阈值的预测结果视为正类（1）
# 计算新预测结果的准确率和召回率
import numpy as np
y_pro=lr.predict_proba(x_test) ##获取预测概率值
y_prd2 = [list(p>=0.02).index(1) for i,p in enumerate(y_pro)]   ##设定0.3阈值，把大于0.3的看成1分类。
train_score=accuracy_score(y_test,y_prd2)
test_recall=recall_score(y_test,y_prd2,average='macro')
print(train_score,test_recall)

print('训练集召回率：',train_recall)
print('测试集召回率：',test_recall)
print('训练集准确率：',train_score)
print('测试集准确率：',test_score)

0.9325356649792913 0.5
训练集召回率： 0.5
测试集召回率： 0.5
训练集准确率： 0.9325356649792913
测试集准确率： 0.9324436263230557

推荐阅读：机器学习指标