Deep Learning Survival Analysis for Consumer Credit Risk Modelling

Explore deep learning survival analysis techniques for consumer credit risk modeling in this 31-minute conference talk by Jiahang Zhong, PhD at ODSC Europe 2019. Gain insights into the importance of accurate credit risk prediction at the individual level for consumer lending and credit card businesses. Learn about the evolution from traditional probabilistic outcome predictions to more precise time-to-event estimates using survival analysis. Discover how recent advancements in big data and deep learning have enabled sophisticated survival models for individual-level predictions. Review theoretical concepts of survival analysis and classic models before delving into the integration of deep learning models in the context of consumer lending. Examine topics such as credit risk scorecards, types of supervised learning, classic survival models, survival in the machine learning era, deep learning survival models, censorship assumptions, and competing hazard objective functions. Enhance your understanding of cutting-edge approaches to credit risk modeling and their potential impact on providing better products to customers and establishing competitive advantages in the market.

Intro
Credit Risk of Personal Loans
Credit Risk Scorecard
Types of supervised learning
Survival analysis
Classic Survival Models
Survival in ML era
Deep Learning Survival Models
Predictions
Censorship assumption
Competing hazard objective function
Competing hazard model