-
Notifications
You must be signed in to change notification settings - Fork 23
/
score.py
476 lines (330 loc) · 13.1 KB
/
score.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
import pandas as pd
import numpy as np
from sklearn.ensemble import RandomForestRegressor
import matplotlib.pyplot as plt
import seaborn as sns
from scipy import stats
import copy
#为了在jupyternotebook画图中显示中文
plt.rcParams['font.sans-serif']=['SimHei']
plt.rcParams['axes.unicode_minus'] = False
#导入数据
train_data = pd.read_csv('data/cs-training.csv')
train_data = train_data.iloc[:,1:]
train_data.info()
#缺失值处理，利用随机森林算法进行填充
mData = train_data.iloc[:,[5,0,1,2,3,4,6,7,8,9]]# 将缺失列 MonthlyIncome放在最前面,为了后面预测值得取值方便，并且重新赋予一个新的dataframe
#MonthlyIncome 非空
train_known = mData[mData.MonthlyIncome.notnull()].as_matrix()
#MonthlyIncome 缺失
train_unknown = mData[mData.MonthlyIncome.isnull()].as_matrix()
# 利用随机森林预测缺失值
train_X = train_known[:,1:]
train_y = train_known[:,0] # 0 列为MonthlyIncome缺失列
rfr=RandomForestRegressor(random_state=0,n_estimators=200,max_depth=3,n_jobs=-1)
rfr.fit(train_X,train_y)
predicted_y = rfr.predict(train_unknown[:,1:]).round(0) #利用预测值填充原始数据中的缺失值
train_data.loc[train_data.MonthlyIncome.isnull(),'MonthlyIncome'] = predicted_y
train_data = train_data.dropna() #舍弃缺失值
train_data = train_data.drop_duplicates() #舍弃重复值
##异常值处理
# 查某些属性的异常值
train_box = train_data.iloc[:,[3,7,9]]
train_box.boxplot()
#去除异常值
train_data=train_data[train_data['NumberOfTime30-59DaysPastDueNotWorse']<90]
train_data=train_data[train_data['age']>0]
#切分数据集
from sklearn.cross_validation import train_test_split
X=train_data.iloc[:,1:]
y=train_data.iloc[:,0]
train_X,test_X,train_y,test_y=train_test_split(X,y,test_size=0.3,random_state=0)
ntrain_data=pd.concat([train_y,train_X],axis=1)
ntest_data = pd.concat([test_y,test_X],axis=1)
#单调分箱
# Y 中 1表示好样本
def mono_bin(Y, X, n=10):
r = 0
good=Y.sum()
bad=Y.count()-good
while np.abs(r) < 1:
d1 = pd.DataFrame({"X": X, "Y": Y, "Bucket": pd.qcut(X, n)})
d2 = d1.groupby('Bucket', as_index = True)
r, p = stats.spearmanr(d2.mean().X, d2.mean().Y)
n = n - 1
d3 = pd.DataFrame(d2.X.min(), columns = ['min'])
d3['min']=d2.min().X
d3['max'] = d2.max().X
d3['sum'] = d2.sum().Y
d3['total'] = d2.count().Y
d3['rate'] = d2.mean().Y
d3['woe']=np.log((d3['rate']/good)/((1-d3['rate'])/bad))
d3['goodattribute']=d3['sum']/good
d3['badattribute']=(d3['total']-d3['sum'])/bad
iv=((d3['goodattribute']-d3['badattribute'])*d3['woe']).sum()
d4 = (d3.sort_index(by = 'min')).reset_index(drop=True)
woe=list(d4['woe'].round(3))
cut=[]
cut.append(float('-inf'))
for i in range(1,n+1):
qua=X.quantile(i/(n+1))
cut.append(round(qua,4))
cut.append(float('inf'))
return d4,iv,cut,woe
#对几个可分箱的属性变量进行分箱操作
x1_d,x1_iv,x1_cut,x1_woe = mono_bin(train_y,train_X.RevolvingUtilizationOfUnsecuredLines)
x2_d,x2_iv,x2_cut,x2_woe = mono_bin(train_y,train_X.age)
x4_d,x4_iv,x4_cut,x4_woe = mono_bin(train_y,train_X.DebtRatio)
x5_d,x5_iv,x5_cut,x5_woe = mono_bin(train_y,train_X.MonthlyIncome)
#对几个有关次数值得属性进行分箱操作--- 离散值分箱
def woe_value(d1):
d2 = d1.groupby('Bucket', as_index = True)
good=train_y.sum()
bad=train_y.count()-good
d3 = pd.DataFrame(d2.X.min(), columns = ['min'])
d3['min']=d2.min().X
d3['max'] = d2.max().X
d3['sum'] = d2.sum().Y
d3['total'] = d2.count().Y
d3['rate'] = d2.mean().Y
d3['woe'] = np.log((d3['rate']/good)/((1-d3['rate'])/bad))
d3['goodattribute']=d3['sum']/good
d3['badattribute']=(d3['total']-d3['sum'])/bad
iv=((d3['goodattribute']-d3['badattribute'])*d3['woe']).sum()
d4 = (d3.sort_index(by = 'min')).reset_index(drop=True)
woe=list(d4['woe'].round(3))
return d4,iv,woe
#特征分箱处理
d3 = pd.DataFrame({"X": train_X['NumberOfTime30-59DaysPastDueNotWorse'], "Y": train_y})
d3['Bucket'] = d3['X']
d3_x1 = d3.loc[(d3['Bucket']<=0)]
d3_x1.loc[:,'Bucket']="(-inf,0]"
d3_x2 = d3.loc[(d3['Bucket']>0) & (d3['Bucket']<= 1)]
d3_x2.loc[:,'Bucket'] = "(0,1]"
d3_x3 = d3.loc[(d3['Bucket']>1) & (d3['Bucket']<= 3)]
d3_x3.loc[:,'Bucket'] = "(1,3]"
d3_x4 = d3.loc[(d3['Bucket']>3) & (d3['Bucket']<= 5)]
d3_x4.loc[:,'Bucket'] = "(3,5]"
d3_x5 = d3.loc[(d3['Bucket']>5)]
d3_x5.loc[:,'Bucket']="(5,+inf)"
d3 = pd.concat([d3_x1,d3_x2,d3_x3,d3_x4,d3_x5])
x3_d,x3_iv,x3_woe= woe_value(d3)
x3_cut = [float('-inf'),0,1,3,5,float('+inf')]
d6 = pd.DataFrame({"X": train_X['NumberOfOpenCreditLinesAndLoans'], "Y": train_y})
d6['Bucket'] = d6['X']
d6_x1 = d6.loc[(d6['Bucket']<=0)]
d6_x1.loc[:,'Bucket']="(-inf,0]"
d6_x2 = d6.loc[(d6['Bucket']>0) & (d6['Bucket']<= 1)]
d6_x2.loc[:,'Bucket'] = "(0,1]"
d6_x3 = d6.loc[(d6['Bucket']>1) & (d6['Bucket']<= 3)]
d6_x3.loc[:,'Bucket'] = "(1,3]"
d6_x4 = d6.loc[(d6['Bucket']>3) & (d6['Bucket']<= 5)]
d6_x4.loc[:,'Bucket'] = "(3,5]"
d6_x5 = d6.loc[(d6['Bucket']>5)]
d6_x5.loc[:,'Bucket']="(5,+inf)"
d6 = pd.concat([d6_x1,d6_x2,d6_x3,d6_x4,d6_x5])
x6_d,x6_iv,x6_woe= woe_value(d6)
x6_cut = [float('-inf'),1,2,3,5,float('+inf')]
d7 = pd.DataFrame({"X": train_X['NumberOfTimes90DaysLate'], "Y": train_y})
d7['Bucket'] = d7['X']
d7_x1 = d7.loc[(d7['Bucket']<=0)]
d7_x1.loc[:,'Bucket']="(-inf,0]"
d7_x2 = d7.loc[(d7['Bucket']>0) & (d7['Bucket']<= 1)]
d7_x2.loc[:,'Bucket'] = "(0,1]"
d7_x3 = d7.loc[(d7['Bucket']>1) & (d7['Bucket']<= 3)]
d7_x3.loc[:,'Bucket'] = "(1,3]"
d7_x4 = d7.loc[(d7['Bucket']>3) & (d7['Bucket']<= 5)]
d7_x4.loc[:,'Bucket'] = "(3,5]"
d7_x5 = d7.loc[(d7['Bucket']>5)]
d7_x5.loc[:,'Bucket']="(5,+inf)"
d7 = pd.concat([d7_x1,d7_x2,d7_x3,d7_x4,d7_x5])
x7_d,x7_iv,x7_woe= woe_value(d7)
x7_cut = [float('-inf'),0,1,3,5,float('+inf')]
d8 = pd.DataFrame({"X": train_X['NumberRealEstateLoansOrLines'], "Y": train_y})
d8['Bucket'] = d8['X']
d8_x1 = d8.loc[(d8['Bucket']<=0)]
d8_x1.loc[:,'Bucket']="(-inf,0]"
d8_x2 = d8.loc[(d8['Bucket']>0) & (d8['Bucket']<= 1)]
d8_x2.loc[:,'Bucket'] = "(0,1]"
d8_x3 = d8.loc[(d8['Bucket']>1) & (d8['Bucket']<= 3)]
d8_x3.loc[:,'Bucket'] = "(1,2]"
d8_x4 = d8.loc[(d8['Bucket']>2) & (d8['Bucket']<= 3)]
d8_x4.loc[:,'Bucket'] = "(2,3]"
d8_x5 = d8.loc[(d8['Bucket']>3)]
d8_x5.loc[:,'Bucket']="(3,+inf)"
d8 = pd.concat([d8_x1,d8_x2,d8_x3,d8_x4,d8_x5])
x8_d,x8_iv,x8_woe= woe_value(d8)
x8_cut = [float('-inf'),0,1,2,3,float('+inf')]
d9 = pd.DataFrame({"X": train_X['NumberOfTime60-89DaysPastDueNotWorse'], "Y": train_y})
d9['Bucket'] = d9['X']
d9_x1 = d9.loc[(d9['Bucket']<=0)]
d9_x1.loc[:,'Bucket']="(-inf,0]"
d9_x2 = d9.loc[(d9['Bucket']>0) & (d9['Bucket']<= 1)]
d9_x2.loc[:,'Bucket'] = "(0,1]"
d9_x3 = d9.loc[(d9['Bucket']>1) & (d9['Bucket']<= 3)]
d9_x3.loc[:,'Bucket'] = "(1,3]"
d9_x4 = d9.loc[(d9['Bucket']>3)]
d9_x4.loc[:,'Bucket']="(3,+inf)"
d9 = pd.concat([d9_x1,d9_x2,d9_x3,d9_x4])
x9_d,x9_iv,x9_woe= woe_value(d9)
x9_cut = [float('-inf'),0,1,3,float('+inf')]
d10 = pd.DataFrame({"X": train_X['NumberOfDependents'], "Y": train_y})
d10['Bucket'] = d10['X']
d10_x1 = d10.loc[(d10['Bucket']<=0)]
d10_x1.loc[:,'Bucket']="(-inf,0]"
d10_x2 = d10.loc[(d10['Bucket']>0) & (d10['Bucket']<= 1)]
d10_x2.loc[:,'Bucket'] = "(0,1]"
d10_x3 = d10.loc[(d10['Bucket']>1) & (d10['Bucket']<= 2)]
d10_x3.loc[:,'Bucket'] = "(1,2]"
d10_x4 = d10.loc[(d10['Bucket']>2) & (d10['Bucket']<= 3)]
d10_x4.loc[:,'Bucket'] = "(2,3]"
d10_x5 = d10.loc[(d10['Bucket']>3) & (d10['Bucket']<= 5)]
d10_x5.loc[:,'Bucket'] = "(3,5]"
d10_x6 = d10.loc[(d10['Bucket']>5)]
d10_x6.loc[:,'Bucket']="(5,+inf)"
d10 = pd.concat([d10_x1,d10_x2,d10_x3,d10_x4,d10_x5,d10_x6])
x10_d,x10_iv,x10_woe= woe_value(d10)
x10_cut = [float('-inf'),0,1,2,3,5,float('+inf')]
#特征筛选：多变量分析---用于剔除特征高度相关的某些特征
corr = train_data.corr()
xticks = ['x0','x1','x2','x3','x4','x5','x6','x7','x8','x9','x10']
yticks = list(corr.index)
fig = plt.figure()
ax1 = fig.add_subplot(1, 1, 1)
sns.heatmap(corr, annot=True, cmap='RdBu', ax=ax1, annot_kws={'size': 5, 'color': 'blue'})
ax1.set_xticklabels(xticks, rotation=0, fontsize=10)
ax1.set_yticklabels(yticks, rotation=0, fontsize=10)
plt.show()
#特征筛选：单变量分析---用于剔除特征高度相关的某些特征
informationValue = []
informationValue.append(x1_iv)
informationValue.append(x2_iv)
informationValue.append(x3_iv)
informationValue.append(x4_iv)
informationValue.append(x5_iv)
informationValue.append(x6_iv)
informationValue.append(x7_iv)
informationValue.append(x8_iv)
informationValue.append(x9_iv)
informationValue.append(x10_iv)
informationValue
index=['x1','x2','x3','x4','x5','x6','x7','x8','x9','x10']
index_num = range(len(index))
ax=plt.bar(index_num,informationValue,tick_label=index)
plt.show()
#基于筛选后的特征，将特征WOE化
def trans_woe(var,var_name,x_woe,x_cut):
woe_name = var_name + '_woe'
for i in range(len(x_woe)):
if i == 0:
var.loc[(var[var_name]<=x_cut[i+1]),woe_name] = x_woe[i]
elif (i>0) and (i<= len(x_woe)-2):
var.loc[((var[var_name]>x_cut[i])&(var[var_name]<=x_cut[i+1])),woe_name] = x_woe[i]
else:
var.loc[(var[var_name]>x_cut[len(x_woe)-1]),woe_name] = x_woe[len(x_woe)-1]
return var
x1_name = 'RevolvingUtilizationOfUnsecuredLines'
x2_name = 'age'
x3_name = 'NumberOfTime30-59DaysPastDueNotWorse'
x7_name = 'NumberOfTimes90DaysLate'
x9_name = 'NumberOfTime60-89DaysPastDueNotWorse'
train_X = trans_woe(train_X,x1_name,x1_woe,x1_cut)
train_X = trans_woe(train_X,x2_name,x2_woe,x2_cut)
train_X = trans_woe(train_X,x3_name,x3_woe,x3_cut)
train_X = trans_woe(train_X,x7_name,x7_woe,x7_cut)
train_X = trans_woe(train_X,x9_name,x9_woe,x9_cut)
#建立模型
## 对训练集结果计算AUC ROC KS值
# # 方法1： 直接用lr预测数值 -- 预测结果0.85
# test_X = trans_woe(test_X,x1_name,x1_woe,x1_cut)
# test_X = trans_woe(test_X,x2_name,x2_woe,x2_cut)
# test_X = trans_woe(test_X,x3_name,x3_woe,x3_cut)
# test_X = trans_woe(test_X,x7_name,x7_woe,x7_cut)
# test_X = trans_woe(test_X,x9_name,x9_woe,x9_cut)
# from sklearn.linear_model.logistic import LogisticRegression
# lr = LogisticRegression()
# lr.fit(train_X, train_y)
# # 注意predict 和predict_proba的区别
# resu = lr.predict_proba(test_X)
# resuLabel = lr.predict(test_X)
# print('predict_proba:',resu)
# print('-------')
# print('predict:',resuLabel)
# from sklearn.metrics import roc_curve,auc
# # X3=sm.add_constant(test_X)
# # resu=result.predict(X3)
# fpr,tpr,thershold=roc_curve(test_y,resu[:,1])
# rocauc=auc(fpr,tpr)
# plt.plot(fpr,tpr,'b',label='AUC=%0.2f'%rocauc)
# plt.legend()
# plt.plot([0,1],[0,1],'r--')
# plt.xlim([0,1])
# plt.ylim([0,1])
# plt.ylabel('TPR')
# plt.xlabel('FPR')
# plt.title('逻辑回归预测')
# plt.show()
# print ('KS:',max(tpr-fpr))
#验证集测试
import statsmodels.api as sm
#为模型增加常数项，即回归线在 y 轴上的截距
X1=sm.add_constant(train_X)
logit=sm.Logit(train_y,X1)
result=logit.fit()
print(result.summary())
test_X = trans_woe(test_X,x1_name,x1_woe,x1_cut)
test_X = trans_woe(test_X,x2_name,x2_woe,x2_cut)
test_X = trans_woe(test_X,x3_name,x3_woe,x3_cut)
test_X = trans_woe(test_X,x7_name,x7_woe,x7_cut)
test_X = trans_woe(test_X,x9_name,x9_woe,x9_cut)
from sklearn.metrics import roc_curve,auc
X3=sm.add_constant(test_X)
resu=result.predict(X3)
fpr,tpr,thershold=roc_curve(test_y,resu)
rocauc=auc(fpr,tpr)
plt.plot(fpr,tpr,'b',label='AUC=%0.2f'%rocauc)
plt.legend()
plt.plot([0,1],[0,1],'r--')
plt.xlim([0,1])
plt.ylim([0,1])
plt.ylabel('TPR')
plt.xlabel('FPR')
plt.show()
#建立评分卡
p = 20/np.log(2) #设置PDO为20
q = 600 - 20*np.log(20)/np.log(2) #设置定点分数为600
def get_score(coe,woe,factor):
scores=[]
for w in woe:
score=round(coe*w*factor,0)
scores.append(score)
return scores
#x_coe表示利用逻辑回归拟合后的系数矩阵
x_coe = [2.6084,0.6327,0.5151,0.5520,0.5747,0.4074]
baseScore = round(q + p * x_coe[0], 0)
x1_score = get_score(x_coe[1], x1_woe, p)
x1_score = get_score(x_coe[1], x1_woe, p)
x2_score = get_score(x_coe[2], x2_woe, p)
x3_score = get_score(x_coe[3], x3_woe, p)
x7_score = get_score(x_coe[4], x7_woe, p)
x9_score = get_score(x_coe[5], x9_woe, p)
score=[x1_score,x2_score,x3_score,x7_score,x9_score]
#建立一个函数使得当输入x1,x2,x3,x7,x9的值时可以返回评分数
cut_t = [x1_cut,x2_cut,x3_cut,x7_cut,x9_cut]
score =[x1_score,x2_score,x3_score,x7_score,x9_score]
def compute_score(x): #x为数组，包含x1,x2,x3,x7和x9的取值
tot_score = baseScore
cut_d = copy.deepcopy(cut_t)
print(cut_d)
for j in range(len(cut_d)):
cut_d[j].append(x[j])
cut_d[j].sort()
print(cut_d[j])
for i in range(len(cut_d[j])):
if cut_d[j][i] == x[j]:
print(score[j][i-1]) #注意一个问题，边界值重复加了两遍，是否需要？
tot_score = score[j][i-1] +tot_score
return tot_score
#输入一个根据筛选后的特征类似的值，并得到最后的评分结果
x_score=[0.3,44,3,3,5]
compute_score(x_score)