风控模型开发验证报告 — DEV-mvgb3cba-b872

00 模型概要信息

项目	内容
模型名称	exp_DEV-mvga8dob-1994
开发完成日期	2026-10-20
模型开发人员	（人工填写）
模型开发人员邮件地址	（人工填写）
模型验证人员	（人工填写）
模型验证人员邮件地址	（人工填写）
模型应用环节	inloan_warning
模型应用频率	（人工填写）
模型开发背景及业务用途描述	Develop run with tuning
模型适用产品/客户描述	（人工填写）
模型重开发原因（如有）	—
模型拟监控频率	（人工填写）
模型路线	LightGBM
目标变量	gbflag
实验 ID	DEV-mvgb3cba-b872

待补充：模型开发/验证人员、应用频率、产品描述等字段需人工补填。

01 模型设计

建模数据样本窗口

项目	内容
切分方式	外部 data_flag 划分（TRAIN 内 6/4 随机开发/测试，VALI 跨时间，PSI 时点）
开发样本(Train)窗口	2021-10-17 ~ 2026-04-30
预留验证样本窗口	2024-04-30 ~ 2026-04-30
跨时间验证样本窗口	2021-12-31 ~ 2026-06-30
PSI时点样本窗口	2023-11-18 ~ 2026-07-31

模型好坏定义

项目	内容
目标变量	gbflag
坏样本(BAD=1)占比	8.03%
好样本(BAD=0)占比	—
不确定样本	本流程要求目标变量已预计算为 0/1 二值，不确定样本请在上游合并口径

样本排除规则

序号	排除规则	说明
1	id_no	主键/时间/口径字段，不入模
2	create_time	主键/时间/口径字段，不入模
3	data_flag	主键/时间/口径字段，不入模
4	gbflag 为空	排除 0 行

滚动率分析

待补充：滚动率分析（M0→M6+ 状态迁移矩阵）需要逾期状态口径数据，当前流程目标变量为上游已计算的二值标签，暂未生成。

待补充：表现窗口及坏账率分析、模型细分定义需结合业务口径人工补充。

02 数据流

开发时点 — 样本排除

类别	样本数
全部样本	115682
排除：目标变量为空	0
排除后样本	115682

注：按照01模型设计的排除规则统计；当前流程的排除项为目标变量缺失。

总体样本概况

数据集	样本量	好样本	坏样本	坏样本率	占比
全部样本	115682	106389	9293	8.03%	100.00%
开发样本	38088	34937	3151	8.27%	32.92%
测试样本	25391	23291	2100	8.27%	21.95%
训练+测试样本	63479	58228	5251	8.27%	54.87%
跨时间验证样本	34730	32298	2432	7.00%	30.02%
PSI时点样本	17473	15863	1610	—	15.10%

开发/预留验证样本交叉统计

类别	开发+预留样本数	开发样本数	开发占比	预留验证样本数	预留验证占比
好	58228	34937	60.00%	23291	40.00%
坏	5251	3151	60.01%	2100	39.99%
合计	63479	38088	60.00%	25391	40.00%

细致变量分析（预分箱口径，开发样本）

变量	IV	缺失率	取值集中度	最大相关系数
due_day_my	1.6852	0.00%	46.32%	0.0972
last_od_to_cur_freq	0.0578	0.02%	70.20%	0.4978
max_od_days	0.0425	0.02%	25.81%	0.5557
last_loan_setl_tnr	0.0385	0.02%	18.96%	0.3034
apprv_amt_day	0.0384	0.02%	79.67%	0.6396
ps_perd_no	0.0351	0.00%	13.10%	0.6425
available_limit_amt	0.0323	0.02%	32.48%	0.8089
end_loan_to_cur_freq	0.0322	0.02%	6.32%	0.6425
last_od_num_m6	0.0313	0.02%	78.72%	0.6476
shd_pre_repay_due_rate	0.0310	1.11%	16.57%	0.5562
odgt0_m12_subcurmonth	0.0306	0.07%	34.86%	0.7415
cre_l60_used_limit_rate_min	0.0306	0.06%	10.21%	0.7274
shd_max_dpd	0.0301	1.05%	74.62%	0.3801
cre_l30_l90_used_amt_d_max	0.0299	0.34%	21.39%	0.6593
setl_prcp_ratio	0.0283	0.02%	21.83%	0.6593
acc_ps_perd_no	0.0277	0.02%	25.81%	0.4281
cre_l90_used_limit_amt_min	0.0261	0.06%	14.62%	0.7390
shd_l30_max_dpd	0.0247	2.96%	71.65%	0.7038
cre_l30_l90_used_amt_m_max	0.0245	0.06%	21.48%	0.6810
loan_os_prcp_credit_amount_m6	0.0243	0.07%	2.98%	0.5204
f_ps_od_days_max	0.0229	0.02%	53.63%	0.5202
apprv_num_m3	0.0228	0.02%	52.89%	0.7533
cre_now_l60_used_amt_m_min	0.0202	0.06%	74.02%	0.6396
loan_os_prcp_credit_amount_m3	0.0202	0.07%	9.66%	0.5204
cre_l60_used_limit_rate_max	0.0178	0.06%	6.42%	0.7274
useratemore50maxcontimonsm12	0.0174	0.07%	10.71%	0.6543
orig_prcp_avg_m3	0.0169	0.02%	60.04%	0.5989
s_ps_od_days_max	0.0166	0.02%	54.12%	0.5455
cre_now_l60_limit_left_d_max	0.0165	0.48%	67.92%	0.7660
od_amt_m6_loan_os_prcp_m6_pc	0.0163	0.07%	93.14%	0.6644
useratemore70_m6_num	0.0159	0.07%	28.42%	0.6941
cre_l30_limit_left_amt_min	0.0149	0.06%	4.18%	0.7875
repay_amount_m3_loan_os_prcp_pc	0.0147	0.07%	38.42%	0.6366
t_ps_od_days_max	0.0144	0.02%	54.18%	0.5455
loan_os_prcp_credit_amount_m12	0.0144	0.07%	0.23%	0.2371
ps_int_amt_mon	0.0141	0.02%	5.50%	0.7503
od_amt_m12_loan_os_prcp_m12_pc	0.0141	0.07%	92.39%	0.6644
suc_apprv_num_m6	0.0139	0.02%	34.85%	0.7214
cre_now_p1m_used_rate_d	0.0136	5.64%	4.54%	0.9733
odgt0_m12_mob_pc	0.0120	0.07%	33.87%	0.6158
useratemore70monthsubcurmonth	0.0120	0.07%	35.85%	0.6077
avg_tnr	0.0117	0.02%	30.34%	0.6841
apprv_num_m1	0.0114	0.02%	73.93%	0.7533
shd_pre_repay_amt_rate	0.0113	0.08%	1.29%	0.6789
cre_now_l30_used_amt_d_min	0.0109	5.71%	86.91%	0.9733
cre_now_l30_used_amt_d_max	0.0108	0.52%	14.27%	0.6000
cre_now_l60_used_amt_m_max	0.0108	0.06%	11.21%	0.6810
min_tnr	0.0106	0.02%	53.18%	0.6841
ps_int_amt_sum	0.0104	0.02%	1.21%	0.7396
suc_apprv_num_m1	0.0104	0.02%	79.23%	0.6800
orig_prcp_avg_m12	0.0103	0.02%	3.24%	0.6695
odgt0_m12_num	0.0102	0.07%	34.86%	0.7141
useratemore90maxcontimonsm12	0.0101	0.07%	24.43%	0.6895
useratemore90_m3_num	0.0097	0.07%	61.13%	0.6941
repay_amount_m6_all_shd_amt_pc	0.0094	0.07%	3.66%	0.7885
cre_l30_l90_usedrate_m_max	0.0088	0.06%	26.85%	0.7107
max_loan_os_prcp_m12	0.0084	0.07%	4.04%	0.7503
m_s_ps_instm_amt	0.0081	0.02%	70.25%	0.4301
cre_now_l60_limit_left_d_min	0.0077	5.74%	11.24%	0.8178
cre_now_l60_usedrate_m_max	0.0076	0.06%	13.38%	0.7107
cre_now_p2m_used_rate_m	0.0075	1.66%	2.23%	0.7553
cre_now_l30_limit_left_d_min	0.0074	4.23%	14.95%	0.3058
last_clear_num_m3	0.0073	0.02%	57.66%	0.7214
ps_int_amt_avg	0.0073	0.02%	1.21%	0.6095
cre_l30_l90_used_amt_m_min	0.0072	0.06%	78.40%	0.5989
cre_now_l60_used_amt_d_min	0.0072	10.25%	81.68%	0.7521
min_apply_amt	0.0067	0.02%	10.57%	0.4938
shd_instm_amt_sum	0.0064	0.08%	0.03%	0.7396
cre_now_l90_limit_left_d_min	0.0060	6.91%	9.77%	0.8178
cre_usedrate_u30_daydelta_max	0.0058	12.49%	83.72%	0.6366
shd_l180_instm_amt_sum	0.0056	0.08%	0.11%	0.7458
cre_now_p3m_used_rate_d	0.0056	11.82%	1.60%	0.7521
repay_amount_m6_loan_os_prcp_pc	0.0055	0.07%	54.31%	0.5912
account_age	0.0053	0.07%	4.74%	0.7206
repay_amount_all_shd_amt_pc	0.0051	0.07%	5.67%	0.6536
odgt0_m3_num	0.0050	0.07%	53.85%	0.7141
shd_pre_repay_cnt	0.0049	0.08%	1.22%	0.5711
shd_l60_min_dpd	0.0047	2.50%	51.07%	0.7466
ps_prcp_amt_mon	0.0042	0.02%	4.60%	0.7458
loan_os_prcp_gt0_m12_num_pc	0.0041	0.07%	61.32%	0.5912
cre_now_p1m_used_rate_m	0.0039	0.93%	4.84%	0.7553
shd_remain_actual_cnt	0.0037	0.08%	7.33%	0.6008
min_apprv_amt	0.0033	0.02%	29.17%	0.5699
max_int_rat	0.0031	0.02%	34.15%	0.4054
account_age_loan_day	0.0031	0.07%	2.11%	0.7206
shd_l90_ps_due_cnt	0.0031	0.08%	17.68%	0.6782
end_loan_to_rep_loan_freq	0.0030	0.02%	6.97%	0.4280
shd_l30_pre_repay_due_rate	0.0029	3.11%	48.56%	0.7855
sum_apply_amt	0.0026	0.02%	10.45%	0.4938
grade_limit_amount	0.0026	0.02%	38.22%	0.4065
shd_l180_max_dpd	0.0021	1.69%	57.75%	0.7885
last_apply_freq	0.0019	0.02%	87.14%	0.6518
setl_rem_prcp_paym_last_month_lo	0.0017	0.07%	87.85%	0.6536
shd_remain_actual_rate	0.0016	0.41%	66.46%	0.7855
shd_min_dpd	0.0016	1.06%	26.93%	0.7113
last_clear_num_m1	0.0015	0.02%	79.20%	0.7141
age	0.0014	0.02%	4.89%	0.1655
cre_now_l30_limit_left_d_max	0.0013	0.72%	79.30%	0.7660
setl_rem_prcp_paym_m12_loan_os_p	0.0013	0.07%	36.61%	0.3691
setl_prcp_mon_ratio	0.0010	0.07%	6.05%	0.4346

注：共 125 个候选变量，按 IV 降序展示前 100 个；集中度 = 最大频次取值样本数 ÷ 非空样本数；相关性为皮尔逊（仅数值变量），完整数据见 variable_detail.json。

03 单变量分析（细分箱）

max_od_days — IV 0.0930，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	9827	8680	1147	11.67%	-0.1437
2	(0.5, 5.5]	21568	20482	1086	5.04%	0.2515
3	(5.5, 7.5]	2317	2114	203	8.76%	-0.0070
4	(7.5, 12.5]	2223	1948	275	12.37%	-0.0552
5	(12.5, inf]	2145	1706	439	20.47%	-0.1527

f_ps_od_days_max — IV 0.0585，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20422	18144	2278	11.15%	-0.1822
2	(0.5, 1.5]	8441	8002	439	5.20%	0.1313
3	(1.5, 2.5]	3420	3275	145	4.24%	0.0838
4	(2.5, 4.5]	3590	3453	137	3.82%	0.0970
5	(4.5, inf]	2207	2056	151	6.84%	0.0197

last_od_to_cur_freq — IV 0.0513，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.5]	4011	3216	795	19.82%	-0.2396
2	(1.5, 3.5]	2608	2372	236	9.05%	-0.0123
3	(3.5, 8.5]	2051	1970	81	3.95%	0.0567
4	(8.5, 50042.0]	2678	2607	71	2.65%	0.0950
5	(50042.0, inf]	26732	24765	1967	7.36%	0.0725

ovd_days_d3_num_m1 — IV 0.0484，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	34676	32228	2448	7.06%	0.1079
2	(0.5, inf]	3404	2702	702	20.62%	-0.2248

odgt0_m12_subcurmonth — IV 0.0471，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	9804	8659	1145	11.68%	-0.1437
2	(0.5, 2.5]	7949	7534	415	5.22%	0.1249
3	(2.5, 5.5]	4022	3860	162	4.03%	0.1018
4	(5.5, 50005.0]	3017	2932	85	2.82%	0.1027
5	(50005.0, inf]	13268	11926	1342	10.11%	-0.0957

end_loan_to_cur_freq — IV 0.0457，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -95.5]	20624	18471	2153	10.44%	-0.1401
2	(-95.5, -68.5]	3011	2762	249	8.27%	0.0000
3	(-68.5, -22.5]	6324	5929	395	6.25%	0.0685
4	(-22.5, -0.5]	5713	5565	148	2.59%	0.1868
5	(-0.5, inf]	2408	2203	205	8.51%	-0.0036

orig_prcp_avg_m3 — IV 0.0438，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 23.449999809265137]	22864	20513	2351	10.28%	-0.1366
2	(23.449999809265137, 1920.6666259765625]	7828	7510	318	4.06%	0.1738
3	(1920.6666259765625, 3515.75]	2938	2790	148	5.04%	0.0584
4	(3515.75, 6174.75]	1980	1853	127	6.41%	0.0233
5	(6174.75, inf]	2470	2264	206	8.34%	-0.0010

t_ps_od_days_max — IV 0.0423，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20633	18517	2116	10.26%	-0.1288
2	(0.5, 2.5]	11097	10524	573	5.16%	0.1613
3	(2.5, 3.5]	2048	1977	71	3.47%	0.0631
4	(3.5, 5.5]	2197	2075	122	5.55%	0.0377
5	(5.5, inf]	2105	1837	268	12.73%	-0.0571

s_ps_od_days_max — IV 0.0405，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20610	18430	2180	10.58%	-0.1484
2	(0.5, 1.5]	7951	7528	423	5.32%	0.1205
3	(1.5, 2.5]	3465	3303	162	4.68%	0.0753
4	(2.5, 4.5]	3494	3326	168	4.81%	0.0730
5	(4.5, inf]	2560	2343	217	8.48%	-0.0032

shd_pre_repay_due_rate — IV 0.0386，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.868800014257431]	10473	9920	553	5.28%	0.1510
2	(0.868800014257431, 0.9498499929904938]	6549	6052	497	7.59%	0.0243
3	(0.9498499929904938, 0.9933499991893768]	5636	5121	515	9.14%	-0.0251
4	(0.9933499991893768, 1.0524500012397766]	9572	8408	1164	12.16%	-0.1599
5	(1.0524500012397766, inf]	5434	5037	397	7.31%	0.0296

last_loan_setl_tnr — IV 0.0380，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	7239	6994	245	3.38%	0.1922
2	(0.5, 3.5]	11453	10542	911	7.95%	0.0158
3	(3.5, 4.5]	2791	2442	349	12.50%	-0.0693
4	(4.5, 8.5]	10684	9521	1163	10.89%	-0.1179
5	(8.5, inf]	5913	5431	482	8.15%	0.0038

available_limit_amt — IV 0.0357，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.03499999921768904]	12385	10918	1467	11.85%	-0.1727
2	(0.03499999921768904, 468.864990234375]	4523	4247	276	6.10%	0.0562
3	(468.864990234375, 4307.300048828125]	11998	11148	850	7.08%	0.0621
4	(4307.300048828125, 27316.265625]	7250	6788	462	6.37%	0.0711
5	(27316.265625, inf]	1924	1829	95	4.94%	0.0410

setl_prcp_ratio — IV 0.0338，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.035099999979138374]	8325	7960	365	4.38%	0.1671
2	(0.035099999979138374, 0.1652499958872795]	4488	4091	397	8.85%	-0.0143
3	(0.1652499958872795, 0.3135499954223633]	4253	3727	526	12.37%	-0.0947
4	(0.3135499954223633, 0.6564500033855438]	12077	10849	1228	10.17%	-0.0933
5	(0.6564500033855438, inf]	8937	8303	634	7.09%	0.0507

cre_l30_l90_used_amt_d_max — IV 0.0332，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.508650004863739]	2891	2707	184	6.36%	0.0340
2	(0.508650004863739, 0.6789500117301941]	8007	7364	643	8.03%	0.0102
3	(0.6789500117301941, 0.8662499785423279]	15190	13563	1627	10.71%	-0.1338
4	(0.8662499785423279, 0.9735499918460846]	3082	2821	261	8.47%	-0.0033
5	(0.9735499918460846, inf]	8790	8358	432	4.91%	0.1496

ps_perd_no — IV 0.0308，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.5]	4991	4845	146	2.93%	0.1562
2	(1.5, 2.5]	3170	2996	174	5.49%	0.0535
3	(2.5, 3.5]	3225	3007	218	6.76%	0.0292
4	(3.5, 9.5]	17375	15601	1774	10.21%	-0.1160
5	(9.5, inf]	9327	8488	839	9.00%	-0.0309

last_od_num_m6 — IV 0.0306，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	29977	27919	2058	6.87%	0.1192
2	(0.5, 3.5]	4179	3621	558	13.35%	-0.1149
3	(3.5, 7.5]	1999	1710	289	14.46%	-0.0751
4	(7.5, inf]	1925	1680	245	12.73%	-0.0527

apprv_num_m3 — IV 0.0302，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20139	18040	2099	10.42%	-0.1376
2	(0.5, 1.5]	6637	6222	415	6.25%	0.0708
3	(1.5, 2.5]	3563	3353	210	5.89%	0.0505
4	(2.5, 4.5]	3610	3431	179	4.96%	0.0717
5	(4.5, inf]	4131	3884	247	5.98%	0.0551

shd_max_dpd — IV 0.0301，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	9565	8427	1138	11.90%	-0.1498
2	(0.5, inf]	28122	26134	1988	7.07%	0.1006

loan_os_prcp_credit_amount_m3 — IV 0.0294，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -4999.0]	3677	3299	378	10.28%	-0.0421
2	(-4999.0, 35.125]	8315	7933	382	4.59%	0.1574
3	(35.125, 100.47500228881836]	19116	17610	1506	7.88%	0.0264
4	(100.47500228881836, 124.34499740600586]	2717	2412	305	11.23%	-0.0476
5	(124.34499740600586, inf]	4235	3657	578	13.65%	-0.1225

suc_apprv_num_m1 — IV 0.0288，3 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	30169	27287	2882	9.55%	-0.0993
2	(0.5, 1.5]	4787	4591	196	4.09%	0.1161
3	(1.5, inf]	3124	3052	72	2.30%	0.1164

max_od_due — IV 0.0279，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	9807	8661	1146	11.69%	-0.1440
2	(0.5, inf]	28253	26250	2003	7.09%	0.0971

apprv_num_m1 — IV 0.0268，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	28152	25425	2727	9.69%	-0.1064
2	(0.5, 1.5]	5350	5107	243	4.54%	0.1130
3	(1.5, 2.5]	1989	1914	75	3.77%	0.0575
4	(2.5, inf]	2589	2484	105	4.06%	0.0684

cre_l60_used_limit_rate_min — IV 0.0264，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.019700000062584877]	4441	4306	135	3.04%	0.1382
2	(0.019700000062584877, 0.09174999967217445]	2813	2656	157	5.58%	0.0466
3	(0.09174999967217445, 0.2787500023841858]	7323	6807	516	7.05%	0.0459
4	(0.2787500023841858, 0.9161500036716461]	18899	17157	1742	9.22%	-0.0603
5	(0.9161500036716461, inf]	4589	3989	600	13.07%	-0.1170

cre_now_l60_used_amt_m_min — IV 0.0256，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.7799999713897705]	28201	25458	2743	9.73%	-0.1091
2	(1.7799999713897705, 869.5950012207031]	2476	2368	108	4.36%	0.0609
3	(869.5950012207031, 1999.7100219726562]	2060	2009	51	2.48%	0.0771
4	(1999.7100219726562, 6462.93505859375]	3125	3009	116	3.71%	0.0880
5	(6462.93505859375, inf]	2203	2071	132	5.99%	0.0316

cre_l30_l90_used_amt_m_max — IV 0.0252，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 35.010000228881836]	8633	8219	414	4.80%	0.1524
2	(35.010000228881836, 350.3800048828125]	6311	5819	492	7.80%	0.0158
3	(350.3800048828125, 1554.1449584960938]	13255	12020	1235	9.32%	-0.0551
4	(1554.1449584960938, 1951.3800048828125]	2059	1896	163	7.92%	0.0046
5	(1951.3800048828125, inf]	7807	6961	846	10.84%	-0.0944

apprv_amt_day — IV 0.0250，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 11.0]	30344	27452	2892	9.53%	-0.0978
2	(11.0, 1000.5]	2506	2421	85	3.39%	0.0773
3	(1000.5, 5995.0]	3304	3220	84	2.54%	0.1172
4	(5995.0, inf]	1926	1837	89	4.62%	0.0450

acc_ps_perd_no — IV 0.0247，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	9827	8680	1147	11.67%	-0.1437
2	(0.5, 7.5]	13221	12111	1110	8.40%	-0.0067
3	(7.5, 23.5]	8461	7896	565	6.68%	0.0665
4	(23.5, 65.5]	4625	4373	252	5.45%	0.0750
5	(65.5, inf]	1946	1870	76	3.91%	0.0546

shd_l30_max_dpd — IV 0.0247，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	27858	25823	2035	7.30%	0.0899
2	(0.5, inf]	9101	8026	1075	11.81%	-0.1374

loan_os_prcp_credit_amount_m6 — IV 0.0239，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 4.129999876022339]	1917	1793	124	6.47%	0.0220
2	(4.129999876022339, 33.795000076293945]	6210	5941	269	4.33%	0.1352
3	(33.795000076293945, 48.94499969482422]	4437	4141	296	6.67%	0.0406
4	(48.94499969482422, 107.59499740600586]	18735	17074	1661	8.87%	-0.0381
5	(107.59499740600586, inf]	6761	5962	799	11.82%	-0.1166

cre_l90_used_limit_amt_min — IV 0.0232，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 21.640000343322754]	5618	5408	210	3.74%	0.1447
2	(21.640000343322754, 748.7099914550781]	8076	7526	550	6.81%	0.0589
3	(748.7099914550781, 8516.22998046875]	20172	18354	1818	9.01%	-0.0490
4	(8516.22998046875, 14597.384765625]	2267	1990	277	12.22%	-0.0541
5	(14597.384765625, inf]	1932	1637	295	15.27%	-0.0821

cre_now_l60_limit_left_d_max — IV 0.0188，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -0.013299999758601189]	2253	1999	254	11.27%	-0.0417
2	(-0.013299999758601189, 0.3223000019788742]	3935	3720	215	5.46%	0.0651
3	(0.3223000019788742, 0.6971500217914581]	2397	2288	109	4.55%	0.0562
4	(0.6971500217914581, 0.9999499917030334]	2890	2809	81	2.80%	0.0993
5	(0.9999499917030334, inf]	26429	23960	2469	9.34%	-0.0810

cre_l30_limit_left_amt_min — IV 0.0169，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -132.05500030517578]	7225	6369	856	11.85%	-0.1230
2	(-132.05500030517578, 101.06500244140625]	4282	4032	250	5.84%	0.0605
3	(101.06500244140625, 432.6999969482422]	3123	2895	228	7.30%	0.0182
4	(432.6999969482422, 2482.1400146484375]	10192	9318	874	8.58%	-0.0137
5	(2482.1400146484375, inf]	13243	12301	942	7.11%	0.0645

avg_od_amt_m6_orig_prcp_pc — IV 0.0166，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5699999928474426]	35568	32866	2702	7.60%	0.0596
2	(0.5699999928474426, inf]	2492	2045	447	17.94%	-0.1391

od_amt_m6_loan_os_prcp_m6_pc — IV 0.0163，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.38500000536441803]	35569	32865	2704	7.60%	0.0591
2	(0.38500000536441803, inf]	2491	2046	445	17.86%	-0.1381

repay_amount_m3_loan_os_prcp_pc — IV 0.0159，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -4999.0]	4804	4554	250	5.20%	0.0845
2	(-4999.0, 31.585000038146973]	5227	4620	607	11.61%	-0.0913
3	(31.585000038146973, 41.34499931335449]	3887	3496	391	10.06%	-0.0393
4	(41.34499931335449, 74.81499862670898]	9434	8576	858	9.09%	-0.0353
5	(74.81499862670898, inf]	14708	13665	1043	7.09%	0.0699

cre_now_l30_limit_left_d_max — IV 0.0155，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.0002500000045984052]	1969	1810	159	8.08%	0.0025
2	(0.0002500000045984052, 0.42524999380111694]	2203	2103	100	4.54%	0.0525
3	(0.42524999380111694, 0.9999499917030334]	3042	2974	68	2.24%	0.1157
4	(0.9999499917030334, inf]	30600	27798	2802	9.16%	-0.0698

cre_now_l60_limit_left_d_min — IV 0.0154，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.04739999957382679]	3632	3344	288	7.93%	0.0107
2	(0.04739999957382679, 0.9717499911785126]	4407	3842	565	12.82%	-0.1076
3	(0.9717499911785126, 1.149049997329712]	9313	8720	593	6.37%	0.0945
4	(1.149049997329712, 3.562299966812134]	15241	13890	1351	8.86%	-0.0263
5	(3.562299966812134, inf]	3308	3083	225	6.80%	0.0331

loan_os_prcp_credit_amount_m12 — IV 0.0153，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 16.18000030517578]	3164	3058	106	3.35%	0.0963
2	(16.18000030517578, 44.21000099182129]	6033	5671	362	6.00%	0.0744
3	(44.21000099182129, 77.875]	14453	13223	1230	8.51%	-0.0134
4	(77.875, 95.76499938964844]	6635	6008	627	9.45%	-0.0394
5	(95.76499938964844, inf]	7775	6951	824	10.60%	-0.0857

cre_now_l90_limit_left_d_min — IV 0.0151，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -0.6798500120639801]	2700	2510	190	7.04%	0.0236
2	(-0.6798500120639801, 0.1424499973654747]	2886	2624	262	9.08%	-0.0123
3	(0.1424499973654747, 0.9937500059604645]	4103	3563	540	13.16%	-0.1111
4	(0.9937500059604645, 1.1872499585151672]	7912	7434	478	6.04%	0.0985
5	(1.1872499585151672, inf]	17855	16347	1508	8.45%	-0.0030

od_amt_m12_loan_os_prcp_m12_pc — IV 0.0143，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.0950000025331974]	35222	32532	2690	7.64%	0.0557
2	(0.0950000025331974, inf]	2838	2379	459	16.17%	-0.1281

shd_l30_pre_repay_due_rate — IV 0.0142，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.8838499784469604]	6969	6144	825	11.84%	-0.1156
2	(0.8838499784469604, 1.0121999979019165]	17955	16588	1367	7.61%	0.0530
3	(1.0121999979019165, 1.781749963760376]	2875	2603	272	9.46%	-0.0180
4	(1.781749963760376, 4.9175498485565186]	7091	6545	546	7.70%	0.0264
5	(4.9175498485565186, inf]	2013	1914	99	4.92%	0.0456

suc_apprv_num_m6 — IV 0.0138，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	13272	11960	1312	9.89%	-0.0843
2	(0.5, 1.5]	8739	7965	774	8.86%	-0.0240
3	(1.5, 2.5]	4770	4438	332	6.96%	0.0351
4	(2.5, 3.5]	3042	2802	240	7.89%	0.0070
5	(3.5, inf]	8257	7765	492	5.96%	0.0960

ps_int_amt_mon — IV 0.0136，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.475000023841858]	2354	2275	79	3.36%	0.0735
2	(1.475000023841858, 19.704999923706055]	9338	8705	633	6.78%	0.0666
3	(19.704999923706055, 132.375]	19413	17827	1586	8.17%	0.0068
4	(132.375, 296.6699981689453]	4803	4267	536	11.16%	-0.0743
5	(296.6699981689453, inf]	2172	1856	316	14.55%	-0.0819

cre_now_l30_used_amt_d_min — IV 0.0132，3 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.0000500082969666]	31211	28273	2938	9.41%	-0.0631
2	(1.0000500082969666, 1.382200002670288]	1933	1860	73	3.78%	0.0612
3	(1.382200002670288, inf]	2768	2687	81	2.93%	0.1006

shd_pre_repay_amt_rate — IV 0.0132，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.6238499879837036]	12637	11814	823	6.51%	0.0964
2	(0.6238499879837036, 0.7506499886512756]	8631	7944	687	7.96%	0.0130
3	(0.7506499886512756, 0.8537499904632568]	9066	8218	848	9.35%	-0.0450
4	(0.8537499904632568, 0.9255499839782715]	5759	5131	628	10.90%	-0.0780
5	(0.9255499839782715, inf]	1966	1803	163	8.29%	-0.0002

cre_l60_used_limit_rate_max — IV 0.0132，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.22405000030994415]	4265	4042	223	5.23%	0.0758
2	(0.22405000030994415, 0.9925500154495239]	20593	18895	1698	8.25%	0.0020
3	(0.9925500154495239, 1.2138000130653381]	5927	5562	365	6.16%	0.0681
4	(1.2138000130653381, 1.424750030040741]	2499	2151	348	13.93%	-0.0834
5	(1.424750030040741, inf]	4781	4265	516	10.79%	-0.0648

cre_now_l30_used_amt_d_max — IV 0.0129，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.6786499917507172]	7376	6898	478	6.48%	0.0691
2	(0.6786499917507172, 0.7513499855995178]	2404	2186	218	9.07%	-0.0112
3	(0.7513499855995178, 0.7632499933242798]	1916	1779	137	7.15%	0.0140
4	(0.7632499933242798, 0.9139499962329865]	17471	15771	1700	9.73%	-0.0866
5	(0.9139499962329865, inf]	8724	8110	614	7.04%	0.0537

useratemore70_m6_num — IV 0.0128，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	10817	10146	671	6.20%	0.1032
2	(0.5, 1.5]	4435	4115	320	7.22%	0.0267
3	(1.5, 4.5]	14412	13104	1308	9.08%	-0.0447
4	(4.5, 5.5]	3468	3132	336	9.69%	-0.0284
5	(5.5, inf]	4928	4414	514	10.43%	-0.0571

odgt0_m12_num — IV 0.0127，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	13268	11926	1342	10.11%	-0.0957
2	(0.5, 1.5]	7737	7100	637	8.23%	0.0015
3	(1.5, 2.5]	5534	5098	436	7.88%	0.0118
4	(2.5, 5.5]	8321	7747	574	6.90%	0.0565
5	(5.5, inf]	3200	3040	160	5.00%	0.0638

useratemore50maxcontimonsm12 — IV 0.0122，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	4076	3889	187	4.59%	0.0889
2	(0.5, 1.5]	2003	1870	133	6.64%	0.0207
3	(1.5, 2.5]	2286	2165	121	5.29%	0.0429
4	(2.5, 4.5]	5893	5401	492	8.35%	-0.0023
5	(4.5, inf]	23802	21586	2216	9.31%	-0.0736

odgt0_m12_mob_pc — IV 0.0121，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 4.164999961853027]	13272	11930	1342	10.11%	-0.0956
2	(4.164999961853027, 20.130000114440918]	12532	11512	1020	8.14%	0.0071
3	(20.130000114440918, 37.14500045776367]	6532	6110	422	6.46%	0.0627
4	(37.14500045776367, 57.670000076293945]	3471	3219	252	7.26%	0.0208
5	(57.670000076293945, inf]	2253	2140	113	5.02%	0.0463

min_tnr — IV 0.0121，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.5]	2688	2434	254	9.45%	-0.0190
2	(1.5, 5.0]	20261	18826	1435	7.08%	0.0837
3	(5.0, 8.5]	3012	2737	275	9.13%	-0.0153
4	(8.5, inf]	12119	10933	1186	9.79%	-0.0752

cre_now_l60_used_amt_d_min — IV 0.0121，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.000950038433075]	27943	25205	2738	9.80%	-0.0698
2	(1.000950038433075, 1.4549500346183777]	2356	2235	121	5.14%	0.0572
3	(1.4549500346183777, 2.305649995803833]	1860	1789	71	3.82%	0.0629
4	(2.305649995803833, inf]	2026	1926	100	4.94%	0.0527

cre_now_l30_limit_left_d_min — IV 0.0114，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.26759999990463257]	2613	2377	236	9.03%	-0.0106
2	(0.26759999990463257, 0.997050017118454]	3523	3066	457	12.97%	-0.0931
3	(0.997050017118454, 1.1242499947547913]	13773	12770	1003	7.28%	0.0632
4	(1.1242499947547913, 1.7257500290870667]	11627	10585	1042	8.96%	-0.0291
5	(1.7257500290870667, inf]	4939	4620	319	6.46%	0.0546

max_loan_os_prcp_m12 — IV 0.0113，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1747.2849731445312]	4101	3860	241	5.88%	0.0574
2	(1747.2849731445312, 2010.25]	2806	2562	244	8.70%	-0.0071
3	(2010.25, 13572.5849609375]	22480	20766	1714	7.62%	0.0473
4	(13572.5849609375, 20848.58984375]	3619	3278	341	9.42%	-0.0239
5	(20848.58984375, inf]	5054	4445	609	12.05%	-0.1001

repay_amount_m6_all_shd_amt_pc — IV 0.0111，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 93.98500061035156]	13257	11946	1311	9.89%	-0.0844
2	(93.98500061035156, 95.59499740600586]	3033	2812	221	7.29%	0.0180
3	(95.59499740600586, 96.48500061035156]	2975	2694	281	9.45%	-0.0207
4	(96.48500061035156, 99.48500061035156]	13735	12687	1048	7.63%	0.0361
5	(99.48500061035156, inf]	5060	4772	288	5.69%	0.0737

setl_rem_prcp_paym_m3_num — IV 0.0110，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	32094	29240	2854	8.89%	-0.0501
2	(0.5, inf]	5966	5671	295	4.94%	0.1096

orig_prcp_avg_m12 — IV 0.0107，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1031.9722290039062]	11498	10735	763	6.64%	0.0841
2	(1031.9722290039062, 1871.875]	6925	6405	520	7.51%	0.0271
3	(1871.875, 3819.857177734375]	8727	7993	734	8.41%	-0.0057
4	(3819.857177734375, 13787.5]	8909	8041	868	9.74%	-0.0603
5	(13787.5, inf]	2021	1756	265	13.11%	-0.0597

avg_tnr — IV 0.0106，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 3.5]	2436	2340	96	3.94%	0.0666
2	(3.5, 5.5]	3672	3442	230	6.26%	0.0436
3	(5.5, 6.5]	2660	2472	188	7.07%	0.0196
4	(6.5, 10.5]	12671	11645	1026	8.10%	0.0092
5	(10.5, inf]	16641	15031	1610	9.67%	-0.0833

ps_int_amt_sum — IV 0.0105，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 209.51499938964844]	3173	3002	171	5.39%	0.0555
2	(209.51499938964844, 815.2900085449219]	10356	9617	739	7.14%	0.0539
3	(815.2900085449219, 1891.7449951171875]	10429	9613	816	7.82%	0.0211
4	(1891.7449951171875, 10533.93994140625]	11316	10276	1040	9.19%	-0.0443
5	(10533.93994140625, inf]	2806	2422	384	13.69%	-0.0883

m_s_ps_instm_amt — IV 0.0103，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 33.55999946594238]	26890	24916	1974	7.34%	0.0741
2	(33.55999946594238, 371.93499755859375]	5271	4739	532	10.09%	-0.0509
3	(371.93499755859375, 614.2250061035156]	2004	1774	230	11.48%	-0.0396
4	(614.2250061035156, 1121.7000122070312]	1960	1785	175	8.93%	-0.0080
5	(1121.7000122070312, inf]	1955	1716	239	12.23%	-0.0476

useratemore70monthsubcurmonth — IV 0.0101，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	13645	12444	1201	8.80%	-0.0287
2	(0.5, 2.5]	6586	5973	613	9.31%	-0.0345
3	(2.5, 4.5]	5198	4685	513	9.87%	-0.0443
4	(4.5, 9.5]	6037	5564	473	7.83%	0.0140
5	(9.5, inf]	6594	6245	349	5.29%	0.1056

cre_now_p1m_used_rate_d — IV 0.0099，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.6786499917507172]	6548	6123	425	6.49%	0.0730
2	(0.6786499917507172, 0.9951499998569489]	21864	19842	2022	9.25%	-0.0466
3	(0.9951499998569489, 1.0000500082969666]	1839	1578	261	14.19%	-0.0649
4	(1.0000500082969666, 1.7942500114440918]	3804	3527	277	7.28%	0.0291
5	(1.7942500114440918, inf]	1885	1790	95	5.04%	0.0436

cre_now_l60_used_amt_m_max — IV 0.0098，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 83.30500030517578]	5148	4904	244	4.74%	0.1035
2	(83.30500030517578, 1001.9599914550781]	16937	15568	1369	8.08%	0.0120
3	(1001.9599914550781, 2003.010009765625]	7151	6515	636	8.89%	-0.0221
4	(2003.010009765625, 2696.469970703125]	2318	2054	264	11.39%	-0.0437
5	(2696.469970703125, inf]	6511	5874	637	9.78%	-0.0496

loan_os_prcp_gt0_m3_num_pc — IV 0.0091，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 68.16500091552734]	3354	3252	102	3.04%	0.1081
2	(68.16500091552734, inf]	34706	31659	3047	8.78%	-0.0423

useratemore90maxcontimonsm12 — IV 0.0090，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	9298	8723	575	6.18%	0.0940
2	(0.5, 1.5]	6903	6349	554	8.03%	0.0087
3	(1.5, 4.5]	14750	13461	1289	8.74%	-0.0265
4	(4.5, 7.5]	4652	4188	464	9.97%	-0.0432
5	(7.5, inf]	2457	2190	267	10.87%	-0.0384

cre_now_p1m_used_rate_m — IV 0.0089，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -0.13974999636411667]	5137	4807	330	6.42%	0.0538
2	(-0.13974999636411667, -4.999999873689376e-05]	2151	2063	88	4.09%	0.0579
3	(-4.999999873689376e-05, 0.0002500000045984052]	1903	1657	246	12.93%	-0.0546
4	(0.0002500000045984052, 0.04154999926686287]	4294	4002	292	6.80%	0.0369
5	(0.04154999926686287, inf]	24249	22076	2173	8.96%	-0.0485

last_clear_num_m3 — IV 0.0087，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	21956	19908	2048	9.33%	-0.0723
2	(0.5, 1.5]	7043	6525	518	7.35%	0.0331
3	(1.5, 2.5]	3417	3177	240	7.02%	0.0253
4	(2.5, 3.5]	2032	1903	129	6.35%	0.0247
5	(3.5, inf]	3632	3417	215	5.92%	0.0507

cre_l30_l90_usedrate_m_max — IV 0.0087，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.025150000117719173]	11879	11080	799	6.73%	0.0811
2	(0.025150000117719173, 0.0799499973654747]	3310	3053	257	7.76%	0.0100
3	(0.0799499973654747, 0.1587500050663948]	6612	5948	664	10.04%	-0.0586
4	(0.1587500050663948, 0.5746999979019165]	14306	13033	1273	8.90%	-0.0347
5	(0.5746999979019165, inf]	1958	1801	157	8.02%	0.0032

cre_l30_l90_used_amt_m_min — IV 0.0087，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.6299999952316284]	29856	27185	2671	8.95%	-0.0528
2	(0.6299999952316284, 1107.5999755859375]	2858	2742	116	4.06%	0.0748
3	(1107.5999755859375, 5082.195068359375]	3258	3082	176	5.40%	0.0566
4	(5082.195068359375, inf]	2093	1906	187	8.93%	-0.0086

useratemore90_m3_num — IV 0.0084，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	23266	21501	1765	7.59%	0.0509
2	(0.5, 1.5]	6117	5646	471	7.70%	0.0185
3	(1.5, 2.5]	3979	3623	356	8.95%	-0.0152
4	(2.5, inf]	4698	4141	557	11.86%	-0.0900

ps_int_amt_avg — IV 0.0082，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 4.333299875259399]	4083	3857	226	5.54%	0.0655
2	(4.333299875259399, 13.080849647521973]	12227	11314	913	7.47%	0.0422
3	(13.080849647521973, 15.02239990234375]	2045	1841	204	9.98%	-0.0216
4	(15.02239990234375, 38.24515151977539]	12306	11265	1041	8.46%	-0.0096
5	(38.24515151977539, inf]	7419	6653	766	10.32%	-0.0736

cre_usedrate_u30_daydelta_max — IV 0.0079，3 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 57.5]	2829	2700	129	4.56%	0.0781
2	(57.5, 85.5]	2112	1992	120	5.68%	0.0435
3	(85.5, inf]	28391	25741	2650	9.33%	-0.0495

min_apply_amt — IV 0.0074，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1135.0]	3388	3010	378	11.16%	-0.0561
2	(1135.0, 2270.0]	5142	4734	408	7.93%	0.0095
3	(2270.0, 4180.0]	12841	11679	1162	9.05%	-0.0406
4	(4180.0, 7300.0]	9221	8490	731	7.93%	0.0149
5	(7300.0, inf]	7488	7017	471	6.29%	0.0761

cre_now_p2m_used_rate_m — IV 0.0071，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -0.11755000054836273]	7939	7380	559	7.04%	0.0504
2	(-0.11755000054836273, -0.00014999999621068127]	2397	2268	129	5.38%	0.0443
3	(-0.00014999999621068127, 0.0799499973654747]	5335	4901	434	8.13%	0.0049
4	(0.0799499973654747, 0.35944999754428864]	17024	15426	1598	9.39%	-0.0659
5	(0.35944999754428864, inf]	4762	4372	390	8.19%	0.0030

cre_now_p3m_used_rate_d — IV 0.0071，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.40674999356269836]	4363	4089	274	6.28%	0.0653
2	(0.40674999356269836, 0.5166500210762024]	4541	4164	377	8.30%	0.0116
3	(0.5166500210762024, 0.8346999883651733]	12866	11590	1276	9.92%	-0.0626
4	(0.8346999883651733, 1.7775499820709229]	8472	7704	768	9.07%	-0.0138
5	(1.7775499820709229, inf]	3343	3104	239	7.15%	0.0335

cre_now_l60_usedrate_m_max — IV 0.0069，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 4.999999873689376e-05]	5094	4749	345	6.77%	0.0425
2	(4.999999873689376e-05, 0.039750000461936]	2632	2499	133	5.05%	0.0527
3	(0.039750000461936, 0.08044999837875366]	3079	2846	233	7.57%	0.0131
4	(0.08044999837875366, 0.5794999897480011]	24469	22245	2224	9.09%	-0.0588
5	(0.5794999897480011, inf]	2791	2576	215	7.70%	0.0097

repay_amount_all_shd_amt_pc — IV 0.0068，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -4999.0]	2159	2017	142	6.58%	0.0230
2	(-4999.0, 90.52499771118164]	9207	8306	901	9.79%	-0.0633
3	(90.52499771118164, 97.21500015258789]	16089	14689	1400	8.70%	-0.0256
4	(97.21500015258789, 98.58499908447266]	6982	6482	500	7.16%	0.0400
5	(98.58499908447266, inf]	3623	3417	206	5.69%	0.0558

account_age — IV 0.0064，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 21.5]	17822	16501	1321	7.41%	0.0562
2	(21.5, 24.5]	3231	2950	281	8.70%	-0.0081
3	(24.5, 29.5]	4283	3947	336	7.85%	0.0104
4	(29.5, 81.5]	9965	8993	972	9.75%	-0.0652
5	(81.5, inf]	2759	2520	239	8.66%	-0.0065

shd_pre_repay_cnt — IV 0.0061，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 25.5]	9560	8876	684	7.15%	0.0504
2	(25.5, 91.5]	11526	10548	978	8.49%	-0.0104
3	(91.5, 237.5]	7531	6977	554	7.36%	0.0348
4	(237.5, 362.5]	2537	2314	223	8.79%	-0.0080
5	(362.5, inf]	6905	6195	710	10.28%	-0.0685

repay_amount_m6_loan_os_prcp_pc — IV 0.0055，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -4999.0]	6828	6353	475	6.96%	0.0467
2	(-4999.0, 51.885000228881836]	1923	1682	241	12.53%	-0.0504
3	(51.885000228881836, 61.88999938964844]	2837	2538	299	10.54%	-0.0381
4	(61.88999938964844, 93.5250015258789]	5789	5260	529	9.14%	-0.0263
5	(93.5250015258789, inf]	20683	19078	1605	7.76%	0.0358

grade_limit_amount — IV 0.0055，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 84500.0]	2178	1945	233	10.70%	-0.0324
2	(84500.0, 115500.0]	9416	8656	760	8.07%	0.0088
3	(115500.0, 137500.0]	2033	1900	133	6.54%	0.0222
4	(137500.0, 180500.0]	6132	5501	631	10.29%	-0.0631
5	(180500.0, inf]	18321	16928	1393	7.60%	0.0440

tothismonthodcontiaddnumm12 — IV 0.0054，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	33057	30463	2594	7.85%	0.0362
2	(0.5, inf]	5003	4448	555	11.09%	-0.0750

shd_l180_instm_amt_sum — IV 0.0052，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 5366.869873046875]	14745	13677	1068	7.24%	0.0608
2	(5366.869873046875, 6463.929931640625]	2051	1854	197	9.61%	-0.0169
3	(6463.929931640625, 34552.490234375]	16659	15283	1376	8.26%	0.0009
4	(34552.490234375, 55994.666015625]	2458	2207	251	10.21%	-0.0289
5	(55994.666015625, inf]	2146	1889	257	11.98%	-0.0484

tothismonthodcontireducenumm12 — IV 0.0050，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	32899	30044	2855	8.68%	-0.0333
2	(0.5, inf]	5161	4867	294	5.70%	0.0747

max_int_rat — IV 0.0048，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.2142000049352646]	3727	3491	236	6.33%	0.0426
2	(0.2142000049352646, 0.2332499995827675]	10809	9880	929	8.59%	-0.0153
3	(0.2332499995827675, 0.23605000227689743]	10068	9109	959	9.53%	-0.0558
4	(0.23605000227689743, inf]	13476	12450	1026	7.61%	0.0365

last_clear_num_m1 — IV 0.0047，3 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	30161	27509	2652	8.79%	-0.0414
2	(0.5, 1.5]	4916	4594	322	6.55%	0.0475
3	(1.5, inf]	3003	2827	176	5.86%	0.0441

shd_l90_ps_due_cnt — IV 0.0046，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 2.5]	2486	2383	103	4.14%	0.0646
2	(2.5, 5.5]	9355	8574	781	8.35%	-0.0032
3	(5.5, 9.5]	8793	7967	826	9.39%	-0.0457
4	(9.5, 17.5]	7082	6546	536	7.57%	0.0255
5	(17.5, inf]	10343	9440	903	8.73%	-0.0210

ps_prcp_amt_mon — IV 0.0046，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 38.48500061035156]	2069	1966	103	4.98%	0.0433
2	(38.48500061035156, 955.4700012207031]	22573	20801	1772	7.85%	0.0306
3	(955.4700012207031, 2836.0150146484375]	9113	8257	856	9.39%	-0.0469
4	(2836.0150146484375, 4656.4951171875]	2267	2085	182	8.03%	0.0034
5	(4656.4951171875, inf]	2058	1821	237	11.52%	-0.0410

shd_instm_amt_sum — IV 0.0045，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 14048.51513671875]	12742	11786	956	7.50%	0.0415
2	(14048.51513671875, 17727.4052734375]	2712	2547	165	6.08%	0.0365
3	(17727.4052734375, 134568.0234375]	17826	16341	1485	8.33%	-0.0036
4	(134568.0234375, 248682.5859375]	2691	2409	282	10.48%	-0.0355
5	(248682.5859375, inf]	2088	1827	261	12.50%	-0.0538

shd_l60_min_dpd — IV 0.0045，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -286.5]	11318	10338	980	8.66%	-0.0135
2	(-286.5, -151.5]	2052	1903	149	7.26%	0.0146
3	(-151.5, -0.5]	1883	1799	84	4.46%	0.0480
4	(-0.5, 0.5]	18965	17424	1541	8.13%	0.0171
5	(0.5, inf]	2917	2558	359	12.31%	-0.0675

shd_remain_actual_cnt — IV 0.0045，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 2.5]	4339	4035	304	7.01%	0.0314
2	(2.5, 3.5]	2790	2627	163	5.84%	0.0417
3	(3.5, 5.5]	4131	3792	339	8.21%	0.0016
4	(5.5, 35.5]	21876	19908	1968	9.00%	-0.0498
5	(35.5, inf]	4923	4548	375	7.62%	0.0179

odgt0_m3_num — IV 0.0044，4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20494	18897	1597	7.79%	0.0333
2	(0.5, 1.5]	9560	8822	738	7.72%	0.0247
3	(1.5, 2.5]	5366	4882	484	9.02%	-0.0214
4	(2.5, inf]	2640	2310	330	12.50%	-0.0660

loan_os_prcp_gt0_m12_num_pc — IV 0.0039，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 57.73500061035156]	3298	3138	160	4.85%	0.0685
2	(57.73500061035156, 72.08000183105469]	2631	2417	214	8.13%	0.0022
3	(72.08000183105469, 76.01499938964844]	1961	1784	177	9.03%	-0.0092
4	(76.01499938964844, 92.31500244140625]	6640	6094	546	8.22%	0.0017
5	(92.31500244140625, inf]	23530	21478	2052	8.72%	-0.0321

account_age_loan_day — IV 0.0039，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 2.603700041770935]	14039	12830	1209	8.61%	-0.0188
2	(2.603700041770935, 6.708349943161011]	9319	8667	652	7.00%	0.0566
3	(6.708349943161011, 10.562049865722656]	3970	3661	309	7.78%	0.0112
4	(10.562049865722656, 25.645400047302246]	5755	5200	555	9.64%	-0.0412
5	(25.645400047302246, inf]	4977	4553	424	8.52%	-0.0067

end_loan_to_rep_loan_freq — IV 0.0037，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1003.5]	28677	26419	2258	7.87%	0.0320
2	(1003.5, 1935.5]	2012	1819	193	9.59%	-0.0165
3	(1935.5, 2132.5]	2266	1984	282	12.44%	-0.0571
4	(2132.5, 2322.5]	2666	2417	249	9.34%	-0.0172
5	(2322.5, inf]	2459	2291	168	6.83%	0.0219

min_apprv_amt — IV 0.0037，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.25]	12574	11446	1128	8.97%	-0.0361
2	(1.25, 127.5]	7484	6970	514	6.87%	0.0534
3	(127.5, 424.5]	4994	4578	416	8.33%	-0.0016
4	(424.5, 1202.0]	6672	6154	518	7.76%	0.0175
5	(1202.0, inf]	6356	5782	574	9.03%	-0.0248

sum_apply_amt — IV 0.0034，5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 4180.0]	18132	16531	1601	8.83%	-0.0353
2	(4180.0, 7925.0]	9886	9087	799	8.08%	0.0086
3	(7925.0, 11214.5]	5589	5226	363	6.49%	0.0544
4	(11214.5, 15038.6298828125]	2480	2290	190	7.66%	0.0093
5	(15038.6298828125, inf]	1993	1796	197	9.88%	-0.0200

last_apply_freq — IV 0.0034，3 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 6.5]	33302	30679	2623	7.88%	0.0336
2	(6.5, 498.5]	2175	1924	251	11.54%	-0.0434
3	(498.5, inf]	2603	2327	276	10.60%	-0.0364

bh_cur_od_num — IV 0.0031，2 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -49999.5]	15040	13682	1358	9.03%	-0.0433
2	(-49999.5, inf]	23040	21248	1792	7.78%	0.0362

注：统计全部候选变量的自动分箱明细（初分组），按 IV 降序展示前 100 个变量，共 124 个；完整数据见 feature_report.json。

04 单变量分析（粗分箱）

f_ps_od_days_max — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20422	18144	2278	11.15%	-0.1822
2	(0.5, 1.5]	8441	8002	439	5.20%	0.1313
3	(1.5, 2.5]	3420	3275	145	4.24%	0.0838
4	(2.5, 4.5]	3590	3453	137	3.82%	0.0970
5	(4.5, inf]	2207	2056	151	6.84%	0.0197

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（11.15% → 11.15% → 6.84%），如需严格单调可考虑调整切点或合并分箱。IV=0.0585，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 4→5 箱有局部反弹：Bad Rate 由 3.82% 回升至 6.84%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 11.15%，可作为拒绝或重点风控策略的阈值参考

t_ps_od_days_max — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20633	18517	2116	10.26%	-0.1288
2	(0.5, 2.5]	11097	10524	573	5.16%	0.1613
3	(2.5, 3.5]	2048	1977	71	3.47%	0.0631
4	(3.5, 5.5]	2197	2075	122	5.55%	0.0377
5	(5.5, inf]	2105	1837	268	12.73%	-0.0571

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.26% → 12.73% → 12.73%），如需严格单调可考虑调整切点或合并分箱。IV=0.0423，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 1→2 箱有局部反弹：Bad Rate 由 10.26% 回落至 5.16%，如需严格单调可考虑合并这两箱

第 5 箱（(5.5, inf]）为坏率最高区间：坏率高达 12.73%，可作为拒绝或重点风控策略的阈值参考

s_ps_od_days_max — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20610	18430	2180	10.58%	-0.1484
2	(0.5, 1.5]	7951	7528	423	5.32%	0.1205
3	(1.5, 2.5]	3465	3303	162	4.68%	0.0753
4	(2.5, 4.5]	3494	3326	168	4.81%	0.0730
5	(4.5, inf]	2560	2343	217	8.48%	-0.0032

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.58% → 10.58% → 8.48%），如需严格单调可考虑调整切点或合并分箱。IV=0.0405，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 3→4 箱有局部反弹：Bad Rate 由 4.68% 回升至 4.81%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 10.58%，可作为拒绝或重点风控策略的阈值参考

available_limit_amt — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.03499999921768904]	12385	10918	1467	11.85%	-0.1727
2	(0.03499999921768904, 468.864990234375]	4523	4247	276	6.10%	0.0562
3	(468.864990234375, 4307.300048828125]	11998	11148	850	7.08%	0.0621
4	(4307.300048828125, 27316.265625]	7250	6788	462	6.37%	0.0711
5	(27316.265625, inf]	1924	1829	95	4.94%	0.0410

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（11.85% → 11.85% → 4.94%），如需严格单调可考虑调整切点或合并分箱。IV=0.0357，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 2→3 箱有局部反弹：Bad Rate 由 6.10% 回升至 7.08%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.03499999921768904]）为坏率最高区间：坏率高达 11.85%，可作为拒绝或重点风控策略的阈值参考

apprv_num_m3 — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	20139	18040	2099	10.42%	-0.1376
2	(0.5, 1.5]	6637	6222	415	6.25%	0.0708
3	(1.5, 2.5]	3563	3353	210	5.89%	0.0505
4	(2.5, 4.5]	3610	3431	179	4.96%	0.0717
5	(4.5, inf]	4131	3884	247	5.98%	0.0551

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.42% → 10.42% → 5.98%），如需严格单调可考虑调整切点或合并分箱。IV=0.0302，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 4→5 箱有局部反弹：Bad Rate 由 4.96% 回升至 5.98%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 10.42%，可作为拒绝或重点风控策略的阈值参考

loan_os_prcp_credit_amount_m3 — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, -4999.0]	3677	3299	378	10.28%	-0.0421
2	(-4999.0, 35.125]	8315	7933	382	4.59%	0.1574
3	(35.125, 100.47500228881836]	19116	17610	1506	7.88%	0.0264
4	(100.47500228881836, 124.34499740600586]	2717	2412	305	11.23%	-0.0476
5	(124.34499740600586, inf]	4235	3657	578	13.65%	-0.1225

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.28% → 13.65% → 13.65%），如需严格单调可考虑调整切点或合并分箱。IV=0.0294，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 1→2 箱有局部反弹：Bad Rate 由 10.28% 回落至 4.59%，如需严格单调可考虑合并这两箱

第 5 箱（(124.34499740600586, inf]）为坏率最高区间：坏率高达 13.65%，可作为拒绝或重点风控策略的阈值参考

suc_apprv_num_m1 — 3 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.5]	30169	27287	2882	9.55%	-0.0993
2	(0.5, 1.5]	4787	4591	196	4.09%	0.1161
3	(1.5, inf]	3124	3052	72	2.30%	0.1164

关键结论

单调性良好：随分栏增大，Bad Rate 单调下降（9.55% → 2.30%），WOE 单调上升（-0.0993 → 0.1164），符合风控分箱预期。IV=0.0288，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 2 箱 WOE<0（风险高于平均）

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 9.55%，可作为拒绝或重点风控策略的阈值参考

cre_l60_used_limit_rate_min — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 0.019700000062584877]	4441	4306	135	3.04%	0.1382
2	(0.019700000062584877, 0.09174999967217445]	2813	2656	157	5.58%	0.0466
3	(0.09174999967217445, 0.2787500023841858]	7323	6807	516	7.05%	0.0459
4	(0.2787500023841858, 0.9161500036716461]	18899	17157	1742	9.22%	-0.0603
5	(0.9161500036716461, inf]	4589	3989	600	13.07%	-0.1170

关键结论

单调性良好：随分栏增大，Bad Rate 单调上升（3.04% → 13.07%），WOE 单调下降（0.1382 → -0.1170），符合风控分箱预期。IV=0.0264，该变量区分度偏弱但可用。

WOE 在第 3、4 箱之间穿越 0 线：前 3 箱 WOE>0（客户质量优于整体），后 2 箱 WOE<0（风险高于平均）

第 5 箱（(0.9161500036716461, inf]）为坏率最高区间：坏率高达 13.07%，可作为拒绝或重点风控策略的阈值参考

cre_now_l60_used_amt_m_min — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 1.7799999713897705]	28201	25458	2743	9.73%	-0.1091
2	(1.7799999713897705, 869.5950012207031]	2476	2368	108	4.36%	0.0609
3	(869.5950012207031, 1999.7100219726562]	2060	2009	51	2.48%	0.0771
4	(1999.7100219726562, 6462.93505859375]	3125	3009	116	3.71%	0.0880
5	(6462.93505859375, inf]	2203	2071	132	5.99%	0.0316

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（9.73% → 9.73% → 5.99%），如需严格单调可考虑调整切点或合并分箱。IV=0.0256，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 3→4 箱有局部反弹：Bad Rate 由 2.48% 回升至 3.71%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 1.7799999713897705]）为坏率最高区间：坏率高达 9.73%，可作为拒绝或重点风控策略的阈值参考

apprv_amt_day — 4 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 11.0]	30344	27452	2892	9.53%	-0.0978
2	(11.0, 1000.5]	2506	2421	85	3.39%	0.0773
3	(1000.5, 5995.0]	3304	3220	84	2.54%	0.1172
4	(5995.0, inf]	1926	1837	89	4.62%	0.0450

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（9.53% → 9.53% → 4.62%），如需严格单调可考虑调整切点或合并分箱。IV=0.0250，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 3 箱 WOE<0（风险高于平均）

第 3→4 箱有局部反弹：Bad Rate 由 2.54% 回升至 4.62%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 11.0]）为坏率最高区间：坏率高达 9.53%，可作为拒绝或重点风控策略的阈值参考

loan_os_prcp_credit_amount_m6 — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 4.129999876022339]	1917	1793	124	6.47%	0.0220
2	(4.129999876022339, 33.795000076293945]	6210	5941	269	4.33%	0.1352
3	(33.795000076293945, 48.94499969482422]	4437	4141	296	6.67%	0.0406
4	(48.94499969482422, 107.59499740600586]	18735	17074	1661	8.87%	-0.0381
5	(107.59499740600586, inf]	6761	5962	799	11.82%	-0.1166

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（6.47% → 11.82% → 11.82%），如需严格单调可考虑调整切点或合并分箱。IV=0.0239，该变量区分度偏弱但可用。

WOE 在第 3、4 箱之间穿越 0 线：前 3 箱 WOE>0（客户质量优于整体），后 2 箱 WOE<0（风险高于平均）

第 1→2 箱有局部反弹：Bad Rate 由 6.47% 回落至 4.33%，如需严格单调可考虑合并这两箱

第 5 箱（(107.59499740600586, inf]）为坏率最高区间：坏率高达 11.82%，可作为拒绝或重点风控策略的阈值参考

cre_l90_used_limit_amt_min — 5 箱

分组号	分栏	Total	Good	Bad	Bad Rate	WOE
1	(-inf, 21.640000343322754]	5618	5408	210	3.74%	0.1447
2	(21.640000343322754, 748.7099914550781]	8076	7526	550	6.81%	0.0589
3	(748.7099914550781, 8516.22998046875]	20172	18354	1818	9.01%	-0.0490
4	(8516.22998046875, 14597.384765625]	2267	1990	277	12.22%	-0.0541
5	(14597.384765625, inf]	1932	1637	295	15.27%	-0.0821

关键结论

单调性良好：随分栏增大，Bad Rate 单调上升（3.74% → 15.27%），WOE 单调下降（0.1447 → -0.0821），符合风控分箱预期。IV=0.0232，该变量区分度偏弱但可用。

WOE 在第 2、3 箱之间穿越 0 线：前 2 箱 WOE>0（客户质量优于整体），后 3 箱 WOE<0（风险高于平均）

第 5 箱（(14597.384765625, inf]）为坏率最高区间：坏率高达 15.27%，可作为拒绝或重点风控策略的阈值参考

注：统计入模变量经合并调整后的最终分箱明细（开发样本口径）。当前流程细/粗分箱由同一自动分箱算法直接产出。

05 变量筛选

项目	内容
筛选优先级	黑名单 > 白名单 > 自动规则
IV 下限	0.02
缺失率上限	0.8
集中度上限	0.95
相关性上限(皮尔逊)	0.7
PSI 上限	0.25
重要性 Top-K	80
黑名单	due_day_my
白名单	无

候选变量 125，成功分箱 124，入模 12

变量	白名单	入模	IV	缺失率	集中度	IV门槛	相关性	PSI	剔除原因
ps_perd_no	—	N	0.0308	PASS	PASS	PASS	—	0.3217	PSI>0.25
due_day_my	—	N	—	—	—	—	—	—	黑名单
age	—	N	0.0010	PASS	PASS	FAIL	—	—	IV<0.02
max_int_rat	—	N	0.0048	PASS	PASS	FAIL	—	—	IV<0.02
min_int_rat	—	N	0.0009	PASS	PASS	FAIL	—	—	IV<0.02
avg_tnr	—	N	0.0106	PASS	PASS	FAIL	—	—	IV<0.02
min_tnr	—	N	0.0121	PASS	PASS	FAIL	—	—	IV<0.02
sum_apply_amt	—	N	0.0034	PASS	PASS	FAIL	—	—	IV<0.02
min_apply_amt	—	N	0.0074	PASS	PASS	FAIL	—	—	IV<0.02
min_apprv_amt	—	N	0.0037	PASS	PASS	FAIL	—	—	IV<0.02
apprv_amt_day	—	Y	0.0250	PASS	PASS	PASS	PASS	0.1356
end_loan_to_rep_loan_freq	—	N	0.0037	PASS	PASS	FAIL	—	—	IV<0.02
end_loan_to_cur_freq	—	N	0.0457	PASS	PASS	PASS	—	0.2531	PSI>0.25
f_rep_loan_to_f_apply_dt_freq	—	N	0.0011	PASS	PASS	FAIL	—	—	IV<0.02
ps_prcp_amt_mon	—	N	0.0046	PASS	PASS	FAIL	—	—	IV<0.02
ps_int_amt_mon	—	N	0.0136	PASS	PASS	FAIL	—	—	IV<0.02
ps_int_amt_sum	—	N	0.0105	PASS	PASS	FAIL	—	—	IV<0.02
ps_prcp_amt_avg	—	N	0.0015	PASS	PASS	FAIL	—	—	IV<0.02
ps_int_amt_avg	—	N	0.0082	PASS	PASS	FAIL	—	—	IV<0.02
setl_prcp_mon_ratio	—	N	0.0024	PASS	PASS	FAIL	—	—	IV<0.02
acc_ps_perd_no	—	N	0.0247	PASS	PASS	PASS	—	0.4588	PSI>0.25
max_od_days	—	N	0.0930	PASS	PASS	PASS	—	0.5635	PSI>0.25
m_s_ps_instm_amt	—	N	0.0103	PASS	PASS	FAIL	—	—	IV<0.02
f_ps_od_days_max	—	Y	0.0585	PASS	PASS	PASS	PASS	0.1507
s_ps_od_days_max	—	Y	0.0405	PASS	PASS	PASS	PASS	0.1952
t_ps_od_days_max	—	Y	0.0423	PASS	PASS	PASS	PASS	0.2430
ovd_days_d3_num_m1	—	N	0.0484	PASS	PASS	PASS	—	0.4567	PSI>0.25
orig_prcp_avg_m3	—	N	0.0438	PASS	PASS	PASS	—	0.2678	PSI>0.25
orig_prcp_avg_m12	—	N	0.0107	PASS	PASS	FAIL	—	—	IV<0.02
suc_apply_trench_num	—	N	0.0016	PASS	PASS	FAIL	—	—	IV<0.02
suc_apply_top_pc_code_num	—	N	0.0011	PASS	PASS	FAIL	—	—	IV<0.02
suc_apprv_num_m1	—	Y	0.0288	PASS	PASS	PASS	PASS	0.1386
apprv_num_m1	—	N	0.0268	PASS	PASS	PASS	FAIL	0.1088	与 apprv_num_m3 相关>0.7
suc_apply_num_m3	—	N	0.0000	PASS	FAIL	—	—	—	集中度>0.95
apprv_num_m3	—	Y	0.0302	PASS	PASS	PASS	PASS	0.2416
suc_apprv_num_m6	—	N	0.0138	PASS	PASS	FAIL	—	—	IV<0.02
apply_num_m12	—	N	0.0002	PASS	PASS	FAIL	—	—	IV<0.02
last_apply_od_num_m12	—	N	0.0015	PASS	PASS	FAIL	—	—	IV<0.02
last_apply_freq	—	N	0.0034	PASS	PASS	FAIL	—	—	IV<0.02
dec_loan_num	—	N	0.0004	PASS	PASS	FAIL	—	—	IV<0.02
bh_cur_od_num	—	N	0.0031	PASS	PASS	FAIL	—	—	IV<0.02
hj_cur_od_num	—	N	0.0000	PASS	PASS	FAIL	—	—	IV<0.02
setl_prcp_ratio	—	N	0.0338	PASS	PASS	PASS	—	0.2983	PSI>0.25
last_loan_setl_tnr	—	N	0.0380	PASS	PASS	PASS	—	0.2903	PSI>0.25
last_od_num_m6	—	N	0.0306	PASS	PASS	PASS	—	0.6013	PSI>0.25
last_od_to_cur_freq	—	N	0.0513	PASS	PASS	PASS	—	0.8376	PSI>0.25
last_clear_num_m1	—	N	0.0047	PASS	PASS	FAIL	—	—	IV<0.02
last_clear_num_m3	—	N	0.0087	PASS	PASS	FAIL	—	—	IV<0.02
grade_limit_amount	—	N	0.0055	PASS	PASS	FAIL	—	—	IV<0.02
exposure	—	N	0.0029	PASS	PASS	FAIL	—	—	IV<0.02
available_limit_amt	—	Y	0.0357	PASS	PASS	PASS	PASS	0.0138
account_age	—	N	0.0064	PASS	PASS	FAIL	—	—	IV<0.02
account_age_loan_day	—	N	0.0039	PASS	PASS	FAIL	—	—	IV<0.02
max_od_due	—	N	0.0279	PASS	PASS	PASS	—	0.4345	PSI>0.25
odgt0_m12_mob_pc	—	N	0.0121	PASS	PASS	FAIL	—	—	IV<0.02
odgt0_m3_num	—	N	0.0044	PASS	PASS	FAIL	—	—	IV<0.02
odgt0_m12_num	—	N	0.0127	PASS	PASS	FAIL	—	—	IV<0.02
odgt0_m12_subcurmonth	—	N	0.0471	PASS	PASS	PASS	—	2.5210	PSI>0.25
tothismonthodcontiaddnumm12	—	N	0.0054	PASS	PASS	FAIL	—	—	IV<0.02
tothismonthodcontireducenumm12	—	N	0.0050	PASS	PASS	FAIL	—	—	IV<0.02
od_amt_m6_loan_os_prcp_m6_pc	—	N	0.0163	PASS	PASS	FAIL	—	—	IV<0.02
od_amt_m12_loan_os_prcp_m12_pc	—	N	0.0143	PASS	PASS	FAIL	—	—	IV<0.02
avg_od_amt_m6_orig_prcp_pc	—	N	0.0166	PASS	PASS	FAIL	—	—	IV<0.02
max_loan_os_prcp_m12	—	N	0.0113	PASS	PASS	FAIL	—	—	IV<0.02
loan_os_prcp_gt0_m3_num_pc	—	N	0.0091	PASS	PASS	FAIL	—	—	IV<0.02
loan_os_prcp_gt0_m12_num_pc	—	N	0.0039	PASS	PASS	FAIL	—	—	IV<0.02
repay_amount_m3_loan_os_prcp_pc	—	N	0.0159	PASS	PASS	FAIL	—	—	IV<0.02
repay_amount_m6_loan_os_prcp_pc	—	N	0.0055	PASS	PASS	FAIL	—	—	IV<0.02
repay_amount_all_shd_amt_pc	—	N	0.0068	PASS	PASS	FAIL	—	—	IV<0.02
repay_amount_m6_all_shd_amt_pc	—	N	0.0111	PASS	PASS	FAIL	—	—	IV<0.02
setl_rem_prcp_paym_last_month_lo	—	N	0.0028	PASS	PASS	FAIL	—	—	IV<0.02
setl_rem_prcp_paym_m12_loan_os_p	—	N	0.0012	PASS	PASS	FAIL	—	—	IV<0.02
setl_rem_prcp_paym_m3_num	—	N	0.0110	PASS	PASS	FAIL	—	—	IV<0.02
setl_rem_prcp_paym_m12_num	—	N	0.0007	PASS	PASS	FAIL	—	—	IV<0.02
loan_os_prcp_credit_amount_m3	—	Y	0.0294	PASS	PASS	PASS	PASS	0.0733
loan_os_prcp_credit_amount_m6	—	Y	0.0239	PASS	PASS	PASS	PASS	0.0433
loan_os_prcp_credit_amount_m12	—	N	0.0153	PASS	PASS	FAIL	—	—	IV<0.02
useratemore50maxcontimonsm12	—	N	0.0122	PASS	PASS	FAIL	—	—	IV<0.02
useratemore70_m6_num	—	N	0.0128	PASS	PASS	FAIL	—	—	IV<0.02
useratemore70monthsubcurmonth	—	N	0.0101	PASS	PASS	FAIL	—	—	IV<0.02
useratemore90_m3_num	—	N	0.0084	PASS	PASS	FAIL	—	—	IV<0.02
useratemore90maxcontimonsm12	—	N	0.0090	PASS	PASS	FAIL	—	—	IV<0.02
shd_max_dpd	—	N	0.0301	PASS	PASS	PASS	—	0.3133	PSI>0.25
shd_min_dpd	—	N	0.0021	PASS	PASS	FAIL	—	—	IV<0.02
shd_pre_repay_cnt	—	N	0.0061	PASS	PASS	FAIL	—	—	IV<0.02
shd_perd_max	—	N	0.0016	PASS	PASS	FAIL	—	—	IV<0.02
shd_instm_amt_sum	—	N	0.0045	PASS	PASS	—	—	—	唯一率过高(ID列)
shd_l30_max_dpd	—	N	0.0247	PASS	PASS	PASS	—	1.1956	PSI>0.25
shd_l60_min_dpd	—	N	0.0045	PASS	PASS	FAIL	—	—	IV<0.02
shd_l90_ps_due_cnt	—	N	0.0046	PASS	PASS	FAIL	—	—	IV<0.02
shd_l180_max_dpd	—	N	0.0021	PASS	PASS	FAIL	—	—	IV<0.02
shd_l180_instm_amt_sum	—	N	0.0052	PASS	PASS	—	—	—	唯一率过高(ID列)
shd_remain_actual_cnt	—	N	0.0045	PASS	PASS	FAIL	—	—	IV<0.02
shd_pre_repay_cnt_rate	—	N	0.0031	PASS	PASS	FAIL	—	—	IV<0.02
shd_pre_repay_amt_rate	—	N	0.0132	PASS	PASS	FAIL	—	—	IV<0.02
shd_pre_repay_30d_amt_rate	—	N	0.0007	PASS	PASS	FAIL	—	—	IV<0.02
shd_pre_repay_due_rate	—	N	0.0386	PASS	PASS	PASS	—	0.3521	PSI>0.25
shd_l30_pre_repay_due_rate	—	N	0.0142	PASS	PASS	FAIL	—	—	IV<0.02
shd_remain_actual_rate	—	N	0.0018	PASS	PASS	FAIL	—	—	IV<0.02
cre_usedrate_u30_daydelta_min	—	N	0.0023	PASS	PASS	FAIL	—	—	IV<0.02
cre_usedrate_u30_daydelta_max	—	N	0.0079	PASS	PASS	FAIL	—	—	IV<0.02
cre_l30_limit_left_amt_min	—	N	0.0169	PASS	PASS	FAIL	—	—	IV<0.02
cre_l60_used_limit_rate_max	—	N	0.0132	PASS	PASS	FAIL	—	—	IV<0.02
cre_l60_used_limit_rate_min	—	Y	0.0264	PASS	PASS	PASS	PASS	0.0410
cre_l90_used_limit_amt_min	—	Y	0.0232	PASS	PASS	PASS	PASS	0.0680
cre_now_p1m_used_rate_d	—	N	0.0099	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_p3m_used_rate_d	—	N	0.0071	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_p1m_used_rate_m	—	N	0.0089	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_p2m_used_rate_m	—	N	0.0071	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l30_used_amt_d_max	—	N	0.0129	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l30_used_amt_d_min	—	N	0.0132	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l30_limit_left_d_max	—	N	0.0155	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l30_limit_left_d_min	—	N	0.0114	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l60_used_amt_d_min	—	N	0.0121	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l60_limit_left_d_max	—	N	0.0188	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l60_limit_left_d_min	—	N	0.0154	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l60_used_amt_m_max	—	N	0.0098	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l60_used_amt_m_min	—	Y	0.0256	PASS	PASS	PASS	PASS	0.0815
cre_now_l60_usedrate_m_max	—	N	0.0069	PASS	PASS	FAIL	—	—	IV<0.02
cre_now_l90_limit_left_d_min	—	N	0.0151	PASS	PASS	FAIL	—	—	IV<0.02
cre_l30_l90_used_amt_d_max	—	N	0.0332	PASS	PASS	PASS	—	0.3521	PSI>0.25
cre_l30_l90_used_amt_m_max	—	N	0.0252	PASS	PASS	PASS	—	0.3031	PSI>0.25
cre_l30_l90_used_amt_m_min	—	N	0.0087	PASS	PASS	FAIL	—	—	IV<0.02
cre_l30_l90_usedrate_m_max	—	N	0.0087	PASS	PASS	FAIL	—	—	IV<0.02
cre_l30_l90_usedrate_m_min	—	N	0.0015	PASS	PASS	FAIL	—	—	IV<0.02

06 回归结果

待补充：本实验未运行 LR 路线（或未产生回归结果），无回归明细。

07 VIF

待补充：本实验未运行 LR 路线，未计算 VIF/相关系数。

08 最终评分卡

待补充：本实验未生成评分卡（未运行 LR 路线）。

09 模型表现

模型区分度评估（按样本）

模型	样本	KS	AUC	GINI
LightGBM	开发样本	0.4474	0.8018	0.6036
LightGBM	测试样本	0.2990	0.7038	0.4076
LightGBM	训练+测试样本	0.3856	0.7625	0.5250
LightGBM	跨时间验证样本	0.2948	0.6946	0.3892

LightGBM — 4 样本评估表（score_cut 粒度）

开发样本（KS 0.0299，AUC 0.5134）

cuts	cnt	bad	good	bad_rate	population_pct	cum_bad_rate	cum_good_rate
[493, 560)	3809	1175	2634	30.85%	10.00%	37.29%	7.54%
[560, 569)	3809	637	3172	16.72%	10.00%	57.51%	16.62%
[569, 575)	3809	396	3413	10.40%	10.00%	70.07%	26.39%
[575, 582)	3808	338	3470	8.88%	10.00%	80.80%	36.32%
[582, 589)	3809	233	3576	6.12%	10.00%	88.19%	46.56%
[589, 597)	3809	136	3673	3.57%	10.00%	92.51%	57.07%
[597, 605)	3808	123	3685	3.23%	10.00%	96.41%	67.62%
[605, 616)	3809	77	3732	2.02%	10.00%	98.86%	78.30%
[616, 633)	3809	31	3778	0.81%	10.00%	99.84%	89.11%
[633, 707)	3809	5	3804	0.13%	10.00%	100.00%	100.00%

测试样本（KS 0.0225，AUC 0.5058）

cuts	cnt	bad	good	bad_rate	population_pct	cum_bad_rate	cum_good_rate
[499, 561)	2540	492	2048	19.37%	10.00%	23.43%	8.79%
[561, 569)	2539	368	2171	14.49%	10.00%	40.95%	18.11%
[569, 575)	2539	307	2232	12.09%	10.00%	55.57%	27.70%
[575, 582)	2539	230	2309	9.06%	10.00%	66.52%	37.61%
[582, 590)	2539	225	2314	8.86%	10.00%	77.24%	47.55%
[590, 597)	2539	160	2379	6.30%	10.00%	84.86%	57.76%
[597, 606)	2539	133	2406	5.24%	10.00%	91.19%	68.09%
[606, 616)	2539	97	2442	3.82%	10.00%	95.81%	78.58%
[616, 633)	2539	65	2474	2.56%	10.00%	98.90%	89.20%
[633, 709)	2539	23	2516	0.91%	10.00%	100.00%	100.00%

训练+测试样本（KS 0.0159，AUC 0.5080）

cuts	cnt	bad	good	bad_rate	population_pct	cum_bad_rate	cum_good_rate
[493, 560)	6348	1665	4683	26.23%	10.00%	31.71%	8.04%
[560, 569)	6348	1010	5338	15.91%	10.00%	50.94%	17.21%
[569, 575)	6348	697	5651	10.98%	10.00%	64.22%	26.91%
[575, 582)	6348	568	5780	8.95%	10.00%	75.03%	36.84%
[582, 590)	6348	459	5889	7.23%	10.00%	83.77%	46.96%
[590, 597)	6347	292	6055	4.60%	10.00%	89.34%	57.35%
[597, 606)	6348	262	6086	4.13%	10.00%	94.32%	67.81%
[606, 616)	6348	173	6175	2.73%	10.00%	97.62%	78.41%
[616, 633)	6348	97	6251	1.53%	10.00%	99.47%	89.15%
[633, 709)	6348	28	6320	0.44%	10.00%	100.00%	100.00%

跨时间验证样本（KS 0.0148，AUC 0.4980）

cuts	cnt	bad	good	bad_rate	population_pct	cum_bad_rate	cum_good_rate
[511, 561)	3473	586	2887	16.87%	10.00%	24.10%	8.94%
[561, 569)	3473	437	3036	12.58%	10.00%	42.06%	18.34%
[569, 576)	3473	333	3140	9.59%	10.00%	55.76%	28.06%
[576, 584)	3473	280	3193	8.06%	10.00%	67.27%	37.95%
[584, 592)	3473	214	3259	6.16%	10.00%	76.07%	48.04%
[592, 599)	3473	164	3309	4.72%	10.00%	82.81%	58.28%
[599, 608)	3473	144	3329	4.15%	10.00%	88.73%	68.59%
[608, 620)	3473	133	3340	3.83%	10.00%	94.20%	78.93%
[620, 640)	3473	84	3389	2.42%	10.00%	97.66%	89.42%
[640, 710)	3473	57	3416	1.64%	10.00%	100.00%	100.00%

跨时间验证样本结论

风险高度集中在低分尾端——最低 20% 样本集中了全量 42.1% 的坏客户，策略切分空间很大。

① 样本与分箱结构：共 34,730 样本 / 2,432 坏样本，整体坏率 7.00%，好坏比约 13.3:1；10 档等频分数分栏，每箱约 10%，分箱均匀，无空箱无过厚箱。

② 排序性与单调性：逐箱 Bad Rate 16.87% → … → 1.64%（低分 → 高分），无倒挂，最低分段坏率是最高分段的 10.3 倍。

③ 区分度与坏客户捕获：最低 20% 样本捕获全量 42.1% 的坏客户，最低 50% 捕获 76.1%；最高 20% 样本仅含 5.8% 的坏客户，其中风险最高一箱单独占 24.1%。

（口径：分数 = 按基准分/优比/PDO 刻度由模型概率转换，分数越高风险越低；结论由程序按分箱数据自动生成。）

注：score_cut 表按分数等频分 10 档（低分 → 高分）：分数由模型概率按基准分/优比/PDO 刻度转换，分数越高风险越低；每表附 KS/AUC（按概率计算，与分栏口径无关）；训练+测试 = TRAIN 全体。旧实验无分数转换时为 PD 等频口径。

10 KS和GINI

LightGBM

样本	KS	GINI	AUC
开发样本	0.4474	0.6036	0.8018
测试样本	0.2990	0.4076	0.7038
训练+测试样本	0.3856	0.5250	0.7625
跨时间验证样本	0.2948	0.3892	0.6946

开发样本

测试样本

训练+测试样本

跨时间验证样本

注：GINI = 2×AUC−1。KS 图 X 轴为分数（升序），红虚线为最大间隔位置；ROC 图例标注 AUC。

11 PSI

模型稳定性评估

模型	训练集与测试集	训练集与跨时间验证	训练集与oot验证集	训练集与psi验证集
LightGBM	0.0003	0.0120	—	0.0077

分数段 PSI 明细（开发样本 vs 跨时间验证样本）

分数段	开发样本占比	跨时间验证样本占比	PSI 贡献	累计 PSI
(-inf,0.0158]	10.00%	13.06%	0.0082	0.0082
(0.0158,0.0284]	10.00%	10.60%	0.0003	0.0085
(0.0284,0.0397]	10.00%	9.31%	0.0005	0.0090
(0.0397,0.0527]	10.00%	10.26%	0.0001	0.0091
(0.0527,0.0673]	10.00%	9.53%	0.0002	0.0093
(0.0673,0.0849]	10.00%	9.01%	0.0010	0.0103
(0.0849,0.1063]	10.00%	9.26%	0.0006	0.0109
(0.1063,0.1292]	10.00%	10.26%	0.0001	0.0110
(0.1292,0.1653]	10.00%	9.62%	0.0001	0.0111
(0.1653,inf]	10.00%	9.09%	0.0009	0.0120

分数段 PSI 明细（开发样本 vs PSI时点样本）

分数段	开发样本占比	PSI时点占比	PSI 贡献	累计 PSI
(-inf,0.0158]	10.00%	11.78%	0.0029	0.0029
(0.0158,0.0284]	10.00%	10.89%	0.0008	0.0037
(0.0284,0.0397]	10.00%	8.86%	0.0014	0.0051
(0.0397,0.0527]	10.00%	10.08%	0.0000	0.0051
(0.0527,0.0673]	10.00%	10.08%	0.0000	0.0051
(0.0673,0.0849]	10.00%	9.26%	0.0006	0.0057
(0.0849,0.1063]	10.00%	8.65%	0.0019	0.0076
(0.1063,0.1292]	10.00%	10.15%	0.0000	0.0076
(0.1292,0.1653]	10.00%	10.33%	0.0001	0.0077
(0.1653,inf]	10.00%	9.91%	0.0000	0.0077

模型样本量分布评估

评分段	开发样本样本量	开发样本占比	测试样本样本量	测试样本占比	训练+测试样本样本量	训练+测试样本占比	跨时间验证样本样本量	跨时间验证样本占比	PSI时点样本样本量	PSI时点样本占比
(-inf, inf]	3809	10.00%	2529	9.96%	6338	9.98%	4536	13.06%	2059	11.78%
(0.0158, 0.0284]	3809	10.00%	2597	10.23%	6406	10.09%	3680	10.60%	1903	10.89%
(0.0284, 0.0397]	3809	10.00%	2514	9.90%	6323	9.96%	3233	9.31%	1548	8.86%
(0.0397, 0.0527]	3808	10.00%	2621	10.32%	6429	10.13%	3562	10.26%	1762	10.08%
(0.0527, 0.0673]	3809	10.00%	2569	10.12%	6378	10.05%	3311	9.53%	1761	10.08%
(0.0673, 0.0849]	3809	10.00%	2480	9.77%	6289	9.91%	3128	9.01%	1618	9.26%
(0.0849, 0.1063]	3808	10.00%	2540	10.00%	6348	10.00%	3217	9.26%	1512	8.65%
(0.1063, 0.1292]	3809	10.00%	2536	9.99%	6345	10.00%	3565	10.26%	1773	10.15%
(0.1292, 0.1653]	3809	10.00%	2512	9.89%	6321	9.96%	3340	9.62%	1805	10.33%
(0.1653, inf]	3809	10.00%	2493	9.82%	6302	9.93%	3158	9.09%	1732	9.91%
sum	38088	100.00%	25391	100.00%	63479	100.00%	34730	100.00%	17473	100.00%

变量 PSI（开发样本 vs PSI时点样本，按入模变量分箱）

特征项	PSI	状态
t_ps_od_days_max	0.2430	关注
apprv_num_m3	0.2416	关注
s_ps_od_days_max	0.1952	关注
f_ps_od_days_max	0.1507	关注
suc_apprv_num_m1	0.1386	关注
apprv_amt_day	0.1356	关注
cre_now_l60_used_amt_m_min	0.0815	稳定
loan_os_prcp_credit_amount_m3	0.0733	稳定
cre_l90_used_limit_amt_min	0.0680	稳定
loan_os_prcp_credit_amount_m6	0.0433	稳定
cre_l60_used_limit_rate_min	0.0410	稳定
available_limit_amt	0.0138	稳定

f_ps_od_days_max — PSI 0.1507

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.5]	53.63%	71.67%	18.04%	0.0523
(0.5, 1.5]	22.17%	15.68%	-6.49%	0.0225
(1.5, 2.5]	8.98%	5.01%	-3.97%	0.0232
(2.5, 4.5]	9.43%	4.97%	-4.45%	0.0285
(4.5, inf]	5.80%	2.67%	-3.12%	0.0242

t_ps_od_days_max — PSI 0.2430

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.5]	54.18%	75.91%	21.72%	0.0732
(0.5, 2.5]	29.14%	17.99%	-11.15%	0.0538
(2.5, 3.5]	5.38%	2.30%	-3.08%	0.0261
(3.5, 5.5]	5.77%	2.56%	-3.21%	0.0261
(5.5, inf]	5.53%	1.25%	-4.28%	0.0637

s_ps_od_days_max — PSI 0.1952

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.5]	54.12%	74.07%	19.95%	0.0626
(0.5, 1.5]	20.88%	14.30%	-6.58%	0.0249
(1.5, 2.5]	9.10%	4.84%	-4.26%	0.0269
(2.5, 4.5]	9.18%	4.50%	-4.68%	0.0333
(4.5, inf]	6.72%	2.30%	-4.42%	0.0474

available_limit_amt — PSI 0.0138

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.03499999921768904]	32.52%	32.39%	-0.14%	0.0000
(0.03499999921768904, 468.864990234375]	11.88%	9.29%	-2.59%	0.0064
(468.864990234375, 4307.300048828125]	31.51%	30.87%	-0.64%	0.0001
(4307.300048828125, 27316.265625]	19.04%	20.46%	1.42%	0.0010
(27316.265625, inf]	5.05%	6.99%	1.94%	0.0063

apprv_num_m3 — PSI 0.2416

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.5]	52.89%	29.54%	-23.35%	0.1360
(0.5, 1.5]	17.43%	30.19%	12.76%	0.0701
(1.5, 2.5]	9.36%	14.08%	4.72%	0.0193
(2.5, 4.5]	9.48%	13.21%	3.73%	0.0124
(4.5, inf]	10.85%	12.99%	2.14%	0.0038

loan_os_prcp_credit_amount_m3 — PSI 0.0733

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, -4999.0]	9.66%	6.42%	-3.24%	0.0133
(-4999.0, 35.125]	21.85%	26.69%	4.85%	0.0097
(35.125, 100.47500228881836]	50.23%	48.56%	-1.67%	0.0006
(100.47500228881836, 124.34499740600586]	7.14%	2.99%	-4.15%	0.0362
(124.34499740600586, inf]	11.13%	15.35%	4.22%	0.0136

suc_apprv_num_m1 — PSI 0.1386

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.5]	79.23%	62.50%	-16.73%	0.0397
(0.5, 1.5]	12.57%	22.98%	10.41%	0.0628
(1.5, inf]	8.20%	14.52%	6.32%	0.0361

cre_l60_used_limit_rate_min — PSI 0.0410

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 0.019700000062584877]	11.67%	16.73%	5.06%	0.0182
(0.019700000062584877, 0.09174999967217445]	7.39%	8.24%	0.84%	0.0009
(0.09174999967217445, 0.2787500023841858]	19.24%	18.81%	-0.43%	0.0001
(0.2787500023841858, 0.9161500036716461]	49.65%	41.27%	-8.38%	0.0155
(0.9161500036716461, inf]	12.06%	14.96%	2.90%	0.0063

cre_now_l60_used_amt_m_min — PSI 0.0815

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 1.7799999713897705]	74.09%	61.70%	-12.38%	0.0227
(1.7799999713897705, 869.5950012207031]	6.50%	8.10%	1.59%	0.0035
(869.5950012207031, 1999.7100219726562]	5.41%	7.26%	1.84%	0.0054
(1999.7100219726562, 6462.93505859375]	8.21%	11.44%	3.23%	0.0107
(6462.93505859375, inf]	5.79%	11.50%	5.72%	0.0393

apprv_amt_day — PSI 0.1356

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 11.0]	79.68%	63.48%	-16.21%	0.0369
(11.0, 1000.5]	6.58%	11.57%	4.99%	0.0281
(1000.5, 5995.0]	8.68%	14.12%	5.44%	0.0265
(5995.0, inf]	5.06%	10.84%	5.78%	0.0441

loan_os_prcp_credit_amount_m6 — PSI 0.0433

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 4.129999876022339]	5.04%	6.36%	1.32%	0.0031
(4.129999876022339, 33.795000076293945]	16.32%	23.19%	6.88%	0.0242
(33.795000076293945, 48.94499969482422]	11.66%	12.11%	0.45%	0.0002
(48.94499969482422, 107.59499740600586]	49.22%	40.80%	-8.43%	0.0158
(107.59499740600586, inf]	17.76%	17.55%	-0.21%	0.0000

cre_l90_used_limit_amt_min — PSI 0.0680

分栏	开发基准比例	PSI时点样本比例	变更	贡献
(-inf, 21.640000343322754]	14.76%	21.47%	6.71%	0.0251
(21.640000343322754, 748.7099914550781]	21.22%	17.21%	-4.01%	0.0084
(748.7099914550781, 8516.22998046875]	52.99%	45.70%	-7.30%	0.0108
(8516.22998046875, 14597.384765625]	5.96%	6.57%	0.61%	0.0006
(14597.384765625, inf]	5.08%	9.06%	3.99%	0.0231

注：PSI ≤ 0.1 视为稳定，0.1~0.25 需关注，> 0.25 不稳定。

12 单变量分析

注：统计所有入模变量的单变量分析（开发样本口径）。预留验证/总体/OOT 口径的分组明细待扩展。

f_ps_od_days_max

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.5]	20422	53.63%	18144	2278	11.15%	7.96
2	(0.5, 1.5]	8441	75.80%	8002	439	5.20%	18.23
3	(1.5, 2.5]	3420	84.78%	3275	145	4.24%	22.59
4	(2.5, 4.5]	3590	94.20%	3453	137	3.82%	25.20
5	(4.5, inf]	2207	100.00%	2056	151	6.84%	13.62
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（11.15% → 11.15% → 6.84%），如需严格单调可考虑调整切点或合并分箱。IV=0.0585，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 4→5 箱有局部反弹：Bad Rate 由 3.82% 回升至 6.84%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 11.15%，可作为拒绝或重点风控策略的阈值参考

t_ps_od_days_max

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.5]	20633	54.18%	18517	2116	10.26%	8.75
2	(0.5, 2.5]	11097	83.32%	10524	573	5.16%	18.37
3	(2.5, 3.5]	2048	88.70%	1977	71	3.47%	27.85
4	(3.5, 5.5]	2197	94.47%	2075	122	5.55%	17.01
5	(5.5, inf]	2105	100.00%	1837	268	12.73%	6.85
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.26% → 12.73% → 12.73%），如需严格单调可考虑调整切点或合并分箱。IV=0.0423，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 1→2 箱有局部反弹：Bad Rate 由 10.26% 回落至 5.16%，如需严格单调可考虑合并这两箱

第 5 箱（(5.5, inf]）为坏率最高区间：坏率高达 12.73%，可作为拒绝或重点风控策略的阈值参考

s_ps_od_days_max

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.5]	20610	54.12%	18430	2180	10.58%	8.45
2	(0.5, 1.5]	7951	75.00%	7528	423	5.32%	17.80
3	(1.5, 2.5]	3465	84.10%	3303	162	4.68%	20.39
4	(2.5, 4.5]	3494	93.28%	3326	168	4.81%	19.80
5	(4.5, inf]	2560	100.00%	2343	217	8.48%	10.80
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.58% → 10.58% → 8.48%），如需严格单调可考虑调整切点或合并分箱。IV=0.0405，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 3→4 箱有局部反弹：Bad Rate 由 4.68% 回升至 4.81%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 10.58%，可作为拒绝或重点风控策略的阈值参考

available_limit_amt

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.03499999921768904]	12385	32.52%	10918	1467	11.85%	7.44
2	(0.03499999921768904, 468.864990234375]	4523	44.40%	4247	276	6.10%	15.39
3	(468.864990234375, 4307.300048828125]	11998	75.91%	11148	850	7.08%	13.12
4	(4307.300048828125, 27316.265625]	7250	94.95%	6788	462	6.37%	14.69
5	(27316.265625, inf]	1924	100.00%	1829	95	4.94%	19.25
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（11.85% → 11.85% → 4.94%），如需严格单调可考虑调整切点或合并分箱。IV=0.0357，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 2→3 箱有局部反弹：Bad Rate 由 6.10% 回升至 7.08%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.03499999921768904]）为坏率最高区间：坏率高达 11.85%，可作为拒绝或重点风控策略的阈值参考

apprv_num_m3

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.5]	20139	52.89%	18040	2099	10.42%	8.59
2	(0.5, 1.5]	6637	70.32%	6222	415	6.25%	14.99
3	(1.5, 2.5]	3563	79.67%	3353	210	5.89%	15.97
4	(2.5, 4.5]	3610	89.15%	3431	179	4.96%	19.17
5	(4.5, inf]	4131	100.00%	3884	247	5.98%	15.72
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.42% → 10.42% → 5.98%），如需严格单调可考虑调整切点或合并分箱。IV=0.0302，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 4→5 箱有局部反弹：Bad Rate 由 4.96% 回升至 5.98%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 10.42%，可作为拒绝或重点风控策略的阈值参考

loan_os_prcp_credit_amount_m3

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, -4999.0]	3677	9.66%	3299	378	10.28%	8.73
2	(-4999.0, 35.125]	8315	31.51%	7933	382	4.59%	20.77
3	(35.125, 100.47500228881836]	19116	81.73%	17610	1506	7.88%	11.69
4	(100.47500228881836, 124.34499740600586]	2717	88.87%	2412	305	11.23%	7.91
5	(124.34499740600586, inf]	4235	100.00%	3657	578	13.65%	6.33
合计		38060		34911	3149	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（10.28% → 13.65% → 13.65%），如需严格单调可考虑调整切点或合并分箱。IV=0.0294，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 1→2 箱有局部反弹：Bad Rate 由 10.28% 回落至 4.59%，如需严格单调可考虑合并这两箱

第 5 箱（(124.34499740600586, inf]）为坏率最高区间：坏率高达 13.65%，可作为拒绝或重点风控策略的阈值参考

suc_apprv_num_m1

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.5]	30169	79.23%	27287	2882	9.55%	9.47
2	(0.5, 1.5]	4787	91.80%	4591	196	4.09%	23.42
3	(1.5, inf]	3124	100.00%	3052	72	2.30%	42.39
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性良好：随分栏增大，Bad Rate 单调下降（9.55% → 2.30%），WOE 单调上升（-0.0993 → 0.1164），符合风控分箱预期。IV=0.0288，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 2 箱 WOE<0（风险高于平均）

第 1 箱（(-inf, 0.5]）为坏率最高区间：坏率高达 9.55%，可作为拒绝或重点风控策略的阈值参考

cre_l60_used_limit_rate_min

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 0.019700000062584877]	4441	11.67%	4306	135	3.04%	31.90
2	(0.019700000062584877, 0.09174999967217445]	2813	19.06%	2656	157	5.58%	16.92
3	(0.09174999967217445, 0.2787500023841858]	7323	38.30%	6807	516	7.05%	13.19
4	(0.2787500023841858, 0.9161500036716461]	18899	87.94%	17157	1742	9.22%	9.85
5	(0.9161500036716461, inf]	4589	100.00%	3989	600	13.07%	6.65
合计		38065		34915	3150	8.28%	11.08

关键结论

单调性良好：随分栏增大，Bad Rate 单调上升（3.04% → 13.07%），WOE 单调下降（0.1382 → -0.1170），符合风控分箱预期。IV=0.0264，该变量区分度偏弱但可用。

WOE 在第 3、4 箱之间穿越 0 线：前 3 箱 WOE>0（客户质量优于整体），后 2 箱 WOE<0（风险高于平均）

第 5 箱（(0.9161500036716461, inf]）为坏率最高区间：坏率高达 13.07%，可作为拒绝或重点风控策略的阈值参考

cre_now_l60_used_amt_m_min

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 1.7799999713897705]	28201	74.09%	25458	2743	9.73%	9.28
2	(1.7799999713897705, 869.5950012207031]	2476	80.59%	2368	108	4.36%	21.93
3	(869.5950012207031, 1999.7100219726562]	2060	86.00%	2009	51	2.48%	39.39
4	(1999.7100219726562, 6462.93505859375]	3125	94.21%	3009	116	3.71%	25.94
5	(6462.93505859375, inf]	2203	100.00%	2071	132	5.99%	15.69
合计		38065		34915	3150	8.28%	11.08

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（9.73% → 9.73% → 5.99%），如需严格单调可考虑调整切点或合并分箱。IV=0.0256，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 4 箱 WOE<0（风险高于平均）

第 3→4 箱有局部反弹：Bad Rate 由 2.48% 回升至 3.71%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 1.7799999713897705]）为坏率最高区间：坏率高达 9.73%，可作为拒绝或重点风控策略的阈值参考

apprv_amt_day

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 11.0]	30344	79.68%	27452	2892	9.53%	9.49
2	(11.0, 1000.5]	2506	86.27%	2421	85	3.39%	28.48
3	(1000.5, 5995.0]	3304	94.94%	3220	84	2.54%	38.33
4	(5995.0, inf]	1926	100.00%	1837	89	4.62%	20.64
合计		38080		34930	3150	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（9.53% → 9.53% → 4.62%），如需严格单调可考虑调整切点或合并分箱。IV=0.0250，该变量区分度偏弱但可用。

WOE 在第 1、2 箱之间穿越 0 线：前 1 箱 WOE>0（客户质量优于整体），后 3 箱 WOE<0（风险高于平均）

第 3→4 箱有局部反弹：Bad Rate 由 2.54% 回升至 4.62%，如需严格单调可考虑合并这两箱

第 1 箱（(-inf, 11.0]）为坏率最高区间：坏率高达 9.53%，可作为拒绝或重点风控策略的阈值参考

loan_os_prcp_credit_amount_m6

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 4.129999876022339]	1917	5.04%	1793	124	6.47%	14.46
2	(4.129999876022339, 33.795000076293945]	6210	21.35%	5941	269	4.33%	22.09
3	(33.795000076293945, 48.94499969482422]	4437	33.01%	4141	296	6.67%	13.99
4	(48.94499969482422, 107.59499740600586]	18735	82.24%	17074	1661	8.87%	10.28
5	(107.59499740600586, inf]	6761	100.00%	5962	799	11.82%	7.46
合计		38060		34911	3149	8.27%	11.09

关键结论

单调性欠佳：Bad Rate 随分栏存在波动（6.47% → 11.82% → 11.82%），如需严格单调可考虑调整切点或合并分箱。IV=0.0239，该变量区分度偏弱但可用。

WOE 在第 3、4 箱之间穿越 0 线：前 3 箱 WOE>0（客户质量优于整体），后 2 箱 WOE<0（风险高于平均）

第 1→2 箱有局部反弹：Bad Rate 由 6.47% 回落至 4.33%，如需严格单调可考虑合并这两箱

第 5 箱（(107.59499740600586, inf]）为坏率最高区间：坏率高达 11.82%，可作为拒绝或重点风控策略的阈值参考

cre_l90_used_limit_amt_min

组号	Value	Tot. Records	累计占比	#Goods	#Bads	Bad Rate	G/B Odds
1	(-inf, 21.640000343322754]	5618	14.76%	5408	210	3.74%	25.75
2	(21.640000343322754, 748.7099914550781]	8076	35.98%	7526	550	6.81%	13.68
3	(748.7099914550781, 8516.22998046875]	20172	88.97%	18354	1818	9.01%	10.10
4	(8516.22998046875, 14597.384765625]	2267	94.92%	1990	277	12.22%	7.18
5	(14597.384765625, inf]	1932	100.00%	1637	295	15.27%	5.55
合计		38065		34915	3150	8.28%	11.08

关键结论

单调性良好：随分栏增大，Bad Rate 单调上升（3.74% → 15.27%），WOE 单调下降（0.1447 → -0.0821），符合风控分箱预期。IV=0.0232，该变量区分度偏弱但可用。

WOE 在第 2、3 箱之间穿越 0 线：前 2 箱 WOE>0（客户质量优于整体），后 3 箱 WOE<0（风险高于平均）

第 5 箱（(14597.384765625, inf]）为坏率最高区间：坏率高达 15.27%，可作为拒绝或重点风控策略的阈值参考

13 压力测试

注：验证压力情境下模型的有效性和区分能力稳定性：a.坏浓度更高样本 b.时间外样本 c.特定特征客群 d.极端时间样本。当前流程自动生成情景1（跨时间样本）与情景2（跨时间样本剔除一半好样本，批注16）。

测试情景1：跨时间验证样本

变量贡献度

变量	贡献度
cre_l90_used_limit_amt_min	896.0000
loan_os_prcp_credit_amount_m6	867.0000
cre_l60_used_limit_rate_min	865.0000
loan_os_prcp_credit_amount_m3	708.0000
available_limit_amt	595.0000
cre_now_l60_used_amt_m_min	347.0000
t_ps_od_days_max	333.0000
apprv_num_m3	333.0000
s_ps_od_days_max	284.0000
f_ps_od_days_max	267.0000

区分度验证 — LightGBM @ 跨时间样本

KS = 0.2948，GINI = 0.3892

测试情景2：跨时间样本剔除一半好样本

指标	数值
样本量	34730 → 18581（剔除好样本 16149）
KS	0.2971
AUC	0.6967
GINI	0.3934

注：情景2 模拟好样本分布右移的压力情形（固定 seed=42，可复现）；剔除后 KS/AUC 下降幅度反映模型对好样本结构的敏感度。