基於python 信用卡評分系統 的數據分析

基於python 信用卡評分系統 的數據分析

import pandas as pd

import matplotlib.pyplot as plt #導入圖像庫

from sklearn.ensemble import RandomForestRegressor

# 用隨機森林對缺失值預測填充函數

def set_missing(df):

    # 把已有的數值型特徵取出來

    process_df = df.ix[:,[5,0,1,2,3,4,6,7,8,9]]

    # 分紅已知該特徵和未知該特徵兩部分

    known = process_df[process_df.MonthlyIncome.notnull()].as_matrix()

    unknown = process_df[process_df.MonthlyIncome.isnull()].as_matrix()

    # X爲特徵屬性值

    X = known[:, 1:]

    # y爲結果標籤值

    y = known[:, 0]

    # fit到RandomForestRegressor之中

    rfr = RandomForestRegressor(random_state=0, n_estimators=200,max_depth=3,n_jobs=-1)

    rfr.fit(X,y)

    # 用獲得的模型進行未知特徵值預測

    predicted = rfr.predict(unknown[:, 1:]).round(0)

    print(predicted)

    # 用獲得的預測結果填補原缺失數據

    df.loc[(df.MonthlyIncome.isnull()), 'MonthlyIncome'] = predicted

    return df

data = pd.read_csv(r'E:\Python\Source\CreditScore\cs-training.csv')

process_df = data.iloc[:,[5,0,1,2,3,4,6,7,8,9]]

known = process_df[process_df.MonthlyIncome.notnull()].as_matrix()

unknown = process_df[process_df.MonthlyIncome.isnull()].as_matrix()

X = known[:, 1:]

y = known[:, 0]

# fit到RandomForestRegressor之中

rfr = RandomForestRegressor(random_state=0, n_estimators=200,max_depth=3,n_jobs=-1)

rfr.fit(X,y)

# 用獲得的模型進行未知特徵值預測

predicted = rfr.predict(unknown[:, 1:]).round(0)

print(predicted)

data.loc[(data.MonthlyIncome.isnull()), 'MonthlyIncome'] = predicted

 

 

 

[8311. 1159. 8311. ... 1159. 2554. 2554.]

data=data.dropna()#刪除比較少的缺失值

data = data.drop_duplicates()#刪除重複項

#異常值處理

#x1 = data["age"]

x2 = data["RevolvingUtilizationOfUnsecuredLines"]

x3 = data["DebtRatio"]

fig = plt.figure(1)

ax = fig.add_subplot(111)

ax.boxplot([x2,x3])

ax.set_xticklabels(["RevolvingUtilizationOfUnsecuredLines","DebtRatio"])

 

Out[48]:

[Text(0,0,'RevolvingUtilizationOfUnsecuredLines'), Text(0,0,'DebtRatio')]

 

#異常值處理

data = data[data["age"] > 0]

data = data[data['NumberOfTime30-59DaysPastDueNotWorse'] < 90]#剔除異常值

# 好壞客戶的總體分析

data['SeriousDlqin2yrs']=1-data['SeriousDlqin2yrs']

grouped = data["SeriousDlqin2yrs"].groupby(data["SeriousDlqin2yrs"]).count()

print("壞客戶佔比：{:.2%}".format(grouped[0]/grouped[1]))

print(grouped)

grouped.plot(kind="bar")

壞客戶佔比：7.16%
SeriousDlqin2yrs
0      9706
1    135648
Name: SeriousDlqin2yrs, dtype: int64

Out[54]:

<matplotlib.axes._subplots.AxesSubplot at 0x126eecc0>

 

 Y = data['SeriousDlqin2yrs']

 

本文經過對kaggle上的Give Me Some Credit數據的挖掘分析，結合信用評分卡的創建原理，從數據的預處理、變量選擇、建模分析到建立信用評分，建立了一個簡單的信用評分系統。本項目還有許多不足之處，好比分箱應當使用最優分箱或卡方分箱，減小人爲分箱的隨機性，此外模型採用的是邏輯迴歸算法，還能夠多多嘗試其餘模型。

 

 

 

 

 

 

​