Credit Risk Modeling Using Machine Learning : Episode 2

We understand safeguarding your portfolio from current economic shocks is your top priority. That means you require a throughout data-driven approach for credit-risk assessment. Then your credit-decision model should be the reflection of the same and needs to be continually fined tuned. Altair with Deep Credit Risk brings Credit Risk Trends and Analytics, a three-part webinar series on credit risk analytics with a focus on the pain points of developing models that reflect the driving economics. In part one webinar, we understood the challenges in the credit risk industry. Build an understanding of the role of liquidity, equity, and many other key banking features. You learned to do feature engineering and efficient selection, for instance, the impact of income and expense shocks on loan repayments. You were able to pre-process credit information, import information, and merge data from various sources. Now, with the right data, in this part-two webinar, you will learn to build credit risk models through machine learning techniques and understand the impact of the economic downturn on the model outcome. You will develop a good understanding of the pitfalls of machine learning and the merits relative to econometric regression techniques. You will be able to predict defaults, payoffs, loss rates, and exposures. You will be able to efficiently validate credit risk models. We will compare and interpret various validation strategies and learn how to build meaningful user interfaces, compare outputs for models in terms of stability, discrimination, and calibration and efficiently communicate outputs. This will give you the confidence to build credit models accurately as we cover deep domain knowledge on- Key Discussion Points: § Non-Linearities of risk drivers § Impact on Illiquidity and Leverage § Machine Learning Models § Bias-Variance trade-off § Predictability and pain-points § Showcase of lender internal data