การพัฒนาแบบจำลองความเสี่ยงด้านเครดิต (Credit Scoring) สำหรับการพิจารณาอนุมัติสินเชื่อ เพื่อจัดการความเสี่ยงในการให้บริการสินเชื่อ โดย สมสวัสดิ์ เตชพลฤทธินันท์
The First NIDA Business Analytics and Data Sciences Conference
3. • Consider behavior / credit of
customers (based on his/her
knowledge or experience)
4. What is a credit scoring ?
When you apply for credit, lenders want to make sure you can comfortably afford to
manage any new borrowing. To do so, they usually calculate a credit score, weighing
up all the relevant information at their disposal - this helps them to assess the
chances that you will be able to repay what you owe.
People with a high score are usually seen as lower risk, and could therefore be more
likely to be granted credit - and possibly at better rates.
--Experian--
5.
6. Type of Credit Scoring
Application Score
Behavior Score
Collection Score
7. • Expert judgment
• Create without historical data
• Suitable for new segment/product
The ways to develop credit scoring
Judgmental
• Historical base
• Statistics technic eg. Decision tree , Logistic
regression , Neural network
• Suitable for product has sufficient data
Statistical
• Combination of expert judgment and statistical
• Suitable for new product/segment which have
similar feature with existing product
Hybrid