What loans and tornadoes have in common? Challenges and use cases of ML in the NPL sector – Lukas Norman & Edgar Bahilo Rodríguez, B2Holding


Session Outline

Zero Inflated Date has been present in the literature since 1992. However, it has always been an area commonly studied using classical statistical methods and linear models. In this talk you will learn what tornados and financial portfolios have in common, what are the most common ML use cases in the NPL sector and recent methods to deal with zero inflated data.

Key Takeaways

  • Debt collection and NPL Industry
  • Historical methods to deal with zero inflated data
  • Custom loss functions and ML based methods
  • ZIPX and ZAPX (Zero Inflated Poisson and Zero Augmented Poisson XGBoost)

Add comment