wangchunjie
|
- Professor
- Supervisor of Doctorate Candidates
- Supervisor of Master's Candidates
- Name (Pinyin):wangchunjie
- Date of Birth:1978-09-18
- E-Mail:
- Date of Employment:2004-06-30
- School/Department:长春工业大学
- Administrative Position:院长
- Education Level:With Certificate of Graduation for Doctorate Study
- Business Address:新图书馆614
- Contact Information:0431-85716480
- Degree:Doctoral degree
- Professional Title:Professor
- Status:Employed
- Alma Mater:吉林大学
- Teacher College:数学与统计学院
- Discipline:Other specialties in Statistics
Contact Information
- ZipCode：
- Fax：
- PostalAddress：
- OfficePhone：
- Email：
- Paper Publications
A Model Based on Survival-based Credit Risk Assessment System of SMEs.
Release time:2022-12-14 Hits:
- Affiliation of Author(s):数学与统计学院
- Teaching and Research Group:信息与计算科学教研室
- Journal:ACM International Conference Proceeding Series
- Funded by:国家自然科学基金项目
- Abstract:Assessment of the credit risk of small and medium-sized enterprises (SMEs) based on their strength and reputation is very important for the banks as the basis for credit decisions. We establish a model of credit risk assessment system for SMEs based on Nonparametric Maximum Likelihood Estimation (NPMLE) of survival function which addresses the Case Ⅱ interval-censored data. An empirical analysis of a real data set about 425 Chinese SMEs is carried out. Firstly, the data information of 425 SMEs are preprocessed and three main factors are considered: enterprise strength, development potential, situation of supply and demand relationship of upstream and downstream. Then we extract appropriate features based on those factors. The association between credit risk of SMEs and the extracted features is discussed and variable selection is implemented by Random Forest. What's more, the prediction of the credit risk is carried out by Double Random Forest (DRF), which is a competitive ensemble method in classification and prediction. The predictive accuracy is evaluated by ROC curves and confusion matrices, and the outcomes show that DRF outperforms Support Vector Machine (SVM) and Random Forest (RF). Combining the outcomes of prediction with the features, the intervals into which the repayment time fall are obtained, and hence the data set is Case Ⅱ interval-censored, i.e., the exact time of repayment is unknown and we only know that the repayment time falls into an interval. The issue of risk assessments based on Case Ⅱ interval-censored has received little attentions in the literature. Here the case of prepayment, on-time and late payment are all under discussion. The NPMLE is applied to estimate the probability of repayment varying with time. The survival curves of SMEs with different credit rating are drawn, thus establishing a credit risk assessment system for SMEs.
- Indexed by:Journal paper
- Page Number:245–251
- Translation or Not:no
- Date of Publication:2022-09-19