Loeng - Classification in Credit Risk: Modelling Probability of Default

Luminori riskimudelite vanemjuht, Dmitrij Celov annab loengu krediidiriski klassifitseerimisest. Loeng toimub 9. aprillil kell 10.15 ruumis 2030. Loeng toimub inglise keeles ning üritusele saab registreerida siin: https://forms.gle/MNkESSztH7mnnjjt7.

Loengu kirjeldus (inglise keeles)

One of the key credit risk parameters in Credit Risk is PD – the probability of customers to default on their credit obligations within next 12 months. The objective of credit institutions is to assign the new and existing customers to rating grades based on their observed features (individual) and overall macroeconomic environment (common). This objective requires a proper classification of customers, as an industry standard, applying a logistic regression model developed other sufficiently large historical dataset and often challenged by various machine learning approaches.

The content of the presentation:

• Data preparation and exploratory data analysis
• Logistic regression
• Case-Control sampling
• Feature Selection and Model Development
• Model Performance Comparison
• Challenging modelling approaches