Skip to content

sprasadhpy/Deep-learning-model-for-survival-analysis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Deep-learning-model-for-survival-analysis

  • The International Accounting Standards Board (IASB) revised their accounting standards for financial instruments, such as loans or mortgages, in response to the 2008 financial crisis in the United States and Europe.
  • This resulted in the International Financial Reporting Standard 9, which became effective on or after January 1, 2018. The new standard requires banks to build
    provisions for expected losses in their loan portfolio, using a more forward-looking approach that aims to result in a more timely recognition of credit losses. - -- Previously, impairment losses were only recognized to the extent of objective evidence of impairment, meaning a loss event needed to occur before an impairment loss could be booked.
  • To implement the new accounting rules, banks need to build models that can accurately evaluate a borrower's risk, with the probability of default being a key credit risk parameter.
  • Classification techniques such as logistic regression and decision trees can be used to classify risky and non-risky loans, but survival analysis is needed for more accurate credit risk calculations as it estimates the time it takes for a customer to default.

[Slides]: Deep-learning-model-for-survival-analysis

[Notebook]: DeepSurv-PCHazard-CoxCC

Results:

Pycox[DeepSurv Library]

Survival probabilities

Pycox

Piecewise Constant Hazard (PC-Hazard) model

Survival probabilities

Piecewise Constant Hazard (PC-Hazard) model

Coxx CC method

Survival probabilities

CoxCC

About

Deep learning Model devloped for the FORD GDIA credit analytics team

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published