On February 12, 2025, the seminar, titled “Improving Econometric and Statistical Bases for Credit Risk Assessment in Credit Institutions,” was presented by Olmas Isakov, a final-year PhD student in Econometrics and Statistics and an independent researcher.
Drawing from his ongoing doctoral research, the presenter explored one of the most critical challenges faced by credit institutions—credit risk assessment. His presentation highlighted the importance of adopting advanced econometric and machine learning techniques to evaluate and predict non-performing loans (NPLs), a key indicator of financial vulnerability in lending institutions.
Isakov introduced a dynamic panel data approach to identify the determinants of NPLs and presented a machine learning workflow designed to predict loan delinquency in microfinance institutions (MFIs). The workflow included the application of eight supervised machine learning methods: Logistic Regression, Linear Discriminant Analysis, Quadratic Discriminant Analysis, Classification Tree, Random Forest, K-Nearest Neighbors, Extreme Gradient Boosting, and Neural Networks.
Participants gained valuable insights into how traditional statistical methods and cutting-edge machine learning tools can be integrated to improve credit risk models. The session concluded with an engaging discussion, where attendees explored the practical implications of the research and its potential applications in the financial sector.
After several discussions, his dissertation was recommended to proceed to the next stage of defense/VIVA according to Supreme Attestation committee requirements.
