【博客地址】：https://blog.csdn.net/sunyaowu315
【博客大綱地址】：https://blog.csdn.net/sunyaowu315/article/details/82905347
需要項目數據、代碼資料，請添加qq群：102755159，或留言聯系筆者郵件發送！！！
如果對金融風控、機器學習、數據科學、大數據分析等感興趣的小伙伴，可添加微信交流（郵件中備注，我會附上微信號）
話不多說，直接上代碼。
Python：汽車金融評分卡
# coding: utf-8
# <h1>Table of Contents<span class="tocSkip"></span></h1>
# <div class="toc" style="margin-top: 1em;"><ul class="toc-item"><li><ul class="toc-item"><li><span><a href="#拒絕推斷" data-toc-modified-id="拒絕推斷-0.1"><span class="toc-item-num">0.1 </span>拒絕推斷</a></span><ul class="toc-item"><li><span><a href="#第一步準備數據集：把解釋變量和被解釋變量分開，這是KNN這個函數的要求" data-toc-modified-id="第一步準備數據集：把解釋變量和被解釋變量分開，這是KNN這個函數的要求-0.1.1"><span class="toc-item-num">0.1.1 </span>第一步準備數據集：把解釋變量和被解釋變量分開，這是KNN這個函數的要求</a></span></li><li><span><a href="#第二步：進行缺失值填補和標準化，這也是knn這個函數的要求" data-toc-modified-id="第二步：進行缺失值填補和標準化，這也是knn這個函數的要求-0.1.2"><span class="toc-item-num">0.1.2 </span>第二步：進行缺失值填補和標準化，這也是knn這個函數的要求</a></span></li><li><span><a href="#第三步：建模并預測" data-toc-modified-id="第三步：建模并預測-0.1.3"><span class="toc-item-num">0.1.3 </span>第三步：建模并預測</a></span></li><li><span><a href="#第四步：將審核通過的申請者和未通過的申請者進行合并" data-toc-modified-id="第四步：將審核通過的申請者和未通過的申請者進行合并-0.1.4"><span class="toc-item-num">0.1.4 </span>第四步：將審核通過的申請者和未通過的申請者進行合并</a></span></li></ul></li><li><span><a href="#建立違約預測模型" data-toc-modified-id="建立違約預測模型-0.2"><span class="toc-item-num">0.2 </span>建立違約預測模型</a></span><ul class="toc-item"><li><span><a href="#粗篩變量" data-toc-modified-id="粗篩變量-0.2.1"><span class="toc-item-num">0.2.1 </span>粗篩變量</a></span></li><li><span><a href="#變量細篩與數據清洗" data-toc-modified-id="變量細篩與數據清洗-0.2.2"><span class="toc-item-num">0.2.2 </span>變量細篩與數據清洗</a></span></li><li><span><a href="#變量分箱WOE轉換" data-toc-modified-id="變量分箱WOE轉換-0.2.3"><span class="toc-item-num">0.2.3 </span>變量分箱WOE轉換</a></span></li><li><span><a href="#構造分類模型" data-toc-modified-id="構造分類模型-0.2.4"><span class="toc-item-num">0.2.4 </span>構造分類模型</a></span></li><li><span><a href="#檢驗模型" data-toc-modified-id="檢驗模型-0.2.5"><span class="toc-item-num">0.2.5 </span>檢驗模型</a></span></li><li><span><a href="#評分卡開發" data-toc-modified-id="評分卡開發-0.2.6"><span class="toc-item-num">0.2.6 </span>評分卡開發</a></span></li></ul></li></ul></li></ul></div># In[1]:import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import os#get_ipython().magic('matplotlib inline')# In[2]:os.chdir(r'F:\script\0\script_credit')# In[3]:accepts = pd.read_csv('accepts.csv')
rejects = pd.read_csv('rejects.csv')# In[ ]:'''
#信用風險建模案例
##數據說明：本數據是一份汽車貸款違約 數據
##名稱---中文含義
##application_id---申請者ID
##account_number---帳戶號
##bad_ind---是否違約
##vehicle_year---汽車購買時間
##vehicle_make---汽車制造商
##bankruptcy_ind---曾經破產標識
##tot_derog---五年內信用不良事件數量(比如手機欠費消號)
##tot_tr---全部帳戶數量
##age_oldest_tr---最久賬號存續時間(月)
##tot_open_tr---在使用帳戶數量
##tot_rev_tr---在使用可循環貸款帳戶數量(比如信用卡)
##tot_rev_debt---在使用可循環貸款帳戶余額(比如信用卡欠款)
##tot_rev_line---可循環貸款帳戶限額(信用卡授權額度)
##rev_util---可循環貸款帳戶使用比例(余額/限額)
##fico_score---FICO打分
##purch_price---汽車購買金額(元)
##msrp---建議售價
##down_pyt---分期付款的首次交款
##loan_term---貸款期限(月)
##loan_amt---貸款金額
##ltv---貸款金額/建議售價*100
##tot_income---月均收入(元)
##veh_mileage---行使歷程(Mile)
##used_ind---是否使用
##weight---樣本權重
'''##################################################################################################################
# ## 一、拒絕推斷# ### 第一步準備數據集：把解釋變量和被解釋變量分開，這是KNN這個函數的要求# In[4]:
#取出部分變量用于做KNN：由于KNN算法要求使用連續變量，因此僅選了部分重要的連續變量用于做KNN模型
accepts_x = accepts[["tot_derog","age_oldest_tr","rev_util","fico_score","ltv"]]
# In[5]:accepts_y = accepts['bad_ind']
# In[6]:rejects_x = rejects[["tot_derog","age_oldest_tr","rev_util","fico_score","ltv"]]# In[ ]:
# ### 第二步：進行缺失值填補和標準化，這也是knn這個函數的要求# In[ ]:
#查看一下數據集的信息
rejects_x.info()# In[ ]:accepts_x.info()# In[ ]:# 利用fancyimpute包中的knn方法進行缺失值填補
# 本人平時建模用均值或眾數填補。只是有人問到多重插補的方法，所以給大家演示一下。使用時注意一下兩點：
# 1、多重插補算法切記不要把Y放到待填補的數據集中
# 2、缺失值大于30%的變量建議不要使用多重插補的方法，因為共線性問題會比較嚴重
import fancyimpute as fimp
accepts_x_filled = pd.DataFrame(fimp.KNN(3).complete(accepts_x.as_matrix()))
accepts_x_filled.columns = accepts_x.columnsrejects_x_filled = pd.DataFrame(fimp.KNN(3).complete(rejects_x.as_matrix()))
rejects_x_filled.columns = rejects_x.columns# In[8]:# 標準化數據
from sklearn.preprocessing import Normalizer
accepts_x_norm = pd.DataFrame(Normalizer().fit_transform(accepts_x_filled))
accepts_x_norm.columns = accepts_x_filled.columnsrejects_x_norm = pd.DataFrame(Normalizer().fit_transform(rejects_x_filled))
rejects_x_norm.columns = rejects_x_filled.columns# ### 第三步：建模并預測# In[9]:# 利用knn模型進行預測，做拒絕推斷
from sklearn.neighbors import NearestNeighbors
from sklearn.neighbors import KNeighborsClassifier
neigh = KNeighborsClassifier(n_neighbors=5, weights='distance')
neigh.fit(accepts_x_norm, accepts_y) # In[10]:rejects['bad_ind'] = neigh.predict(rejects_x_norm)# ### 第四步：將審核通過的申請者和未通過的申請者進行合并# In[ ]:# accepts的數據是針對于違約用戶的過度抽樣
#因此，rejects也要進行同樣比例的抽樣# In[11]:rejects_res = rejects[rejects['bad_ind'] == 0].sample(1340)
rejects_res = pd.concat([rejects_res, rejects[rejects['bad_ind'] == 1]], axis = 0)# In[12]:data = pd.concat([accepts.iloc[:, 2:-1], rejects_res.iloc[:,1:]], axis = 0)##################################################################################################################
# ## 二、建立違約預測模型# ### 粗篩變量# In[13]:# 分類變量轉換
bankruptcy_dict = {'N':0, 'Y':1}
data.bankruptcy_ind = data.bankruptcy_ind.map(bankruptcy_dict)# In[14]:# 蓋帽法處理年份變量中的異常值，并將年份其轉化為距現在多長時間
# 此處只是一個示例，所有連續變量都要按此方法進行處理
year_min = data.vehicle_year.quantile(0.1)
year_max = data.vehicle_year.quantile(0.99)
data.vehicle_year = data.vehicle_year.map(lambda x: year_min if x <= year_min else x)
data.vehicle_year = data.vehicle_year.map(lambda x: year_max if x >= year_max else x)data.vehicle_year = data.vehicle_year.map(lambda x: 2018 - x)# In[15]:data.drop(['vehicle_make'], axis = 1, inplace = True)# In[ ]:data_filled = pd.DataFrame(fimp.KNN(3).complete(data.as_matrix()))
data_filled.columns = data.columns# In[17]:X = data_filled[['age_oldest_tr', 'bankruptcy_ind', 'down_pyt', 'fico_score','loan_amt', 'loan_term', 'ltv', 'msrp', 'purch_price', 'rev_util','tot_derog', 'tot_income', 'tot_open_tr', 'tot_rev_debt','tot_rev_line', 'tot_rev_tr', 'tot_tr', 'used_ind', 'veh_mileage','vehicle_year']]
y = data_filled['bad_ind']# In[18]:# 利用隨機森林填補變量
from sklearn.ensemble import RandomForestClassifier
clf = RandomForestClassifier(max_depth=5, random_state=0)
clf.fit(X,y)# In[19]:importances = list(clf.feature_importances_)
importances_order = importances.copy()
importances_order.sort(reverse=True)cols = list(X.columns)
col_top = []
for i in importances_order[:9]:col_top.append((i,cols[importances.index(i)]))
col_top# In[20]:col = [i[1] for i in col_top]# ### 變量細篩與數據清洗# In[21]:from PyWoE import WoE
import warnings
warnings.filterwarnings("ignore")# In[22]:data_filled.head()# In[23]:iv_c = {}
for i in col:try:iv_c[i] = WoE(v_type='c').fit(data_filled[i],data_filled['bad_ind']).optimize().iv except:print(i)pd.Series(iv_c).sort_values(ascending=False)# ### 變量分箱WOE轉換# In[24]:WOE_c = data_filled[col].apply(lambda col:WoE(v_type='c').fit(col,data_filled['bad_ind']).optimize().fit_transform(col,data_filled['bad_ind']))# In[25]:WOE_c.head()# ### 構造分類模型# In[26]:# 劃分數據集
from sklearn.cross_validation import train_test_split
X = WOE_c
y = data_filled['bad_ind']
X_train, X_test, y_train, y_test = train_test_split(X,y,test_size = 0.3, random_state = 0)# In[27]: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)plt.title(title)plt.colorbar()tick_marks = np.arange(len(classes))plt.xticks(tick_marks, classes, rotation=0)plt.yticks(tick_marks, classes)thresh = cm.max() / 2.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')# In[28]:# 構建邏輯回歸模型，進行違約概率預測
import itertools
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix,recall_score,classification_report
lr = LogisticRegression(C = 1, penalty = 'l1')
lr.fit(X_train,y_train.values.ravel())
y_pred = lr.predict(X_test.values)# Compute confusion matrix
cnf_matrix = confusion_matrix(y_test,y_pred)
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]
plt.figure()
plot_confusion_matrix(cnf_matrix, classes=class_names, title='Confusion matrix')
plt.show()# In[46]:## 加入代價敏感參數，重新計算
import itertools
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import confusion_matrix,recall_score,classification_report
lr = LogisticRegression(C = 1, penalty = 'l1', class_weight='balanced')
lr.fit(X_train,y_train.values.ravel())
y_pred = lr.predict(X_test.values)# Compute confusion matrix
cnf_matrix = confusion_matrix(y_test,y_pred)
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]
plt.figure()
plot_confusion_matrix(cnf_matrix, classes=class_names, title='Confusion matrix')
plt.show()# ### 檢驗模型# In[47]:from sklearn.metrics import roc_curve, auc
fpr,tpr,threshold = roc_curve(y_test,y_pred, drop_intermediate=False) ###計算真正率和假正率
roc_auc = auc(fpr,tpr) ###計算auc的值 plt.figure()
lw = 2
plt.figure(figsize=(10,10))
plt.plot(fpr, tpr, color='darkorange', lw=lw, label='ROC curve (area = %0.2f)' % roc_auc) ###假正率為橫坐標，真正率為縱坐標做曲線
plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('Receiver operating characteristic example')
plt.legend(loc="lower right")
plt.show()# In[31]:# 利用sklearn.metrics中的roc_curve算出tpr，fpr作圖fig, ax = plt.subplots()
ax.plot(1 - threshold, tpr, label='tpr') # ks曲線要按照預測概率降序排列，所以需要1-threshold鏡像
ax.plot(1 - threshold, fpr, label='fpr')
ax.plot(1 - threshold, tpr-fpr,label='KS')
plt.xlabel('score')
plt.title('KS Curve')
#plt.xticks(np.arange(0,1,0.2), np.arange(1,0,-0.2))
#plt.xticks(np.arange(0,1,0.2), np.arange(score.max(),score.min(),-0.2*(data['反欺詐評分卡總分'].max() - data['反欺詐評分卡總分'].min())))
plt.figure(figsize=(20,20))
legend = ax.legend(loc='upper left', shadow=True, fontsize='x-large')plt.show()# ### 評分卡開發# In[149]:# 求各變量各水平得分
n = 0
for i in X.columns:if n == 0:temp = WoE(v_type='c').fit(data_filled[i],data_filled['bad_ind']).optimize().binstemp['name'] = [i]*len(temp)scorecard = temp.copy()n += 1else:temp = WoE(v_type='c').fit(data_filled[i],data_filled['bad_ind']).optimize().binstemp['name'] = [i]*len(temp)scorecard = pd.concat([scorecard, temp], axis = 0)n += 1scorecard['score'] = scorecard['woe'].map(lambda x: -int(np.ceil(28.8539*x)))# In[151]:# 基準分
print('base score is {}'.format(int(np.ceil(28.8539*lr.intercept_[0]+513.561))))# In[153]:scorecard# In[154]:# 求原始數據表中每個樣本的得分
def fico_score_cnvnt(x):if x < 6.657176e+02:return -21else:return 16def age_oldest_tr_cnvnt(x):if x < 1.618624e+02:return -9else:return 20def rev_util_cnvnt(x):if x < 7.050000e+01:return 7else:return -19def ltv_cnvnt(x):if x < 9.450000e+01:return 16else:return -8def tot_tr_cnvnt(x):if x < 1.085218e+01:return -13elif x < 1.330865e+01:return -4elif x < 1.798767e+01:return 3else:return 11def tot_rev_line_cnvnt(x):if x < 1.201000e+04:return -12else:return 19def tot_derog_cnvnt(x):if x < 1.072596e+00:return 8else:return -13def purch_price_cnvnt(x):if x < 1.569685e+04:return -5else:return 3def tot_rev_debt_cnvnt(x):if x < 1.024000e+04:return -2else:return 8# In[155]:func = [fico_score_cnvnt,age_oldest_tr_cnvnt,rev_util_cnvnt,ltv_cnvnt,tot_tr_cnvnt,tot_rev_line_cnvnt,tot_derog_cnvnt,purch_price_cnvnt,tot_rev_debt_cnvnt]# In[156]:X_score_dict = {i:j for i,j in zip(X.columns,func)}# In[157]:X_score = data_filled[X.columns].copy()
for i in X_score.columns:X_score[i] = X_score[i].map(X_score_dict[i])# In[158]:X_score['SCORE'] = X_score[X.columns].apply(lambda x: sum(x) + 513, axis = 1)# In[159]:X_score_label = pd.concat([X_score, data_filled['bad_ind']], axis = 1)# In[160]:X_score_label.head()# In[161]:# 查看逾期未逾期評分分布
fig, ax = plt.subplots()
ax1 = sns.kdeplot(X_score_label[X_score_label['bad_ind'] == 1]['SCORE'],label='1')
ax2 = sns.kdeplot(X_score_label[X_score_label['bad_ind'] == 0]['SCORE'],label='0')plt.show()
總結
以上是生活随笔為你收集整理的【项目实战】汽车金融评分卡的全部內容，希望文章能夠幫你解決所遇到的問題。
如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。
