Acceso abierto

Research on Dynamic Prediction Model of Consumer Credit Risk under Fintech Innovation

  
17 mar 2025

Cite
Descargar portada

Figure 1.

ROC curve (AUC value) of several algorithms
ROC curve (AUC value) of several algorithms

Figure 2.

KS curve (KS value) of several algorithms
KS curve (KS value) of several algorithms

Figure 3.

Precision-recall curve of several algorithms
Precision-recall curve of several algorithms

Figure 4.

Cox proportional hazards model survival time prediction result
Cox proportional hazards model survival time prediction result

Model parameters and significance test results

Code Covariable coef exp(coef) se(coef) P
X1 Interest rate 0.05 1.16 0.05 <0.005
X2 Highest valid credits -0.08 0.97 0.07 0.015
X3 M1 and above overdue frequency 0.14 1.18 0.06 <0.005
X4 Housing loan repay the amount of this month 0 1 0.04 0.088
X5 Credit query number in 6 months 0.08 1.09 0.03 <0.005
X6 First overdue 0.42 1.52 0.03 <0.005
X7 First overdue amount 0.03 1.04 0.04 <0.005
X8 Current credit -0.78 0.55 0.09 <0.005

Characteristic variable correlation coefficient

X1 X2 X3 X4 X5 X6 X7 X8
X1 1 -0.055 -0.042 0.018 -0.042 0.035 -0.088 -0.065
X2 -0.055 1 0.452 0.033 0.105 -0.052 0.164 0.273
X3 -0.042 0.452 1 0.138 0.046 -0.037 0.153 0.242
X4 0.018 0.033 0.138 1 -0.114 -0.009 0.033 0.092
X5 -0.042 0.105 0.046 -0.114 1 0.010 0.003 -0.094
X6 0.035 -0.052 -0.037 -0.009 0.010 1 0.221 -0.119
X7 -0.088 0.164 0.153 0.033 0.003 0.221 1 0.379
X8 -0.065 0.273 0.242 0.092 -0.094 -0.119 0.379 1

Covariable PH test results

Code Covariable P
X1 Interest rate 0.21
X2 Highest valid credits 0.88
X3 M1 and above overdue frequency 0.32
X4 Housing loan repay the amount of this month 0.59
X5 Credit query number in 6 months 0.23
X6 First overdue 0.27
X7 First overdue amount 0.24
X8 Current credit 0.69