Research

Research statement

My research focuses on developing trustworthy AI methods for decision-making under uncertainty, with applications in insurance, finance, and other high-stakes domains. I focus on representation learning, uncertainty quantification, and statistically grounded predictive modeling for settings where reliability, transparency, and risk matter. I have several on-going works related to: applying deep neural networks to actuarial rate-making and loss reserving, analyzing unstructured data, and numerically optimizing operational decisions under uncertainty. 

Research themes

1. Trustworthy AI and uncertainty quantification 

I study how predictive models can better account for uncertainty in high-stakes settings, especially when decisions depend on more than point prediction. This includes work on statistically grounded machine learning, probabilistic modeling, and uncertainty-aware representations.

• Lee, G. Y.  and Weaver, J. (Working paper, 2026), A probabilistic representation-learning framework with embedding uncertainty.

• Lee, G. Y. and Li, J. (Working paper, 2026), County-linked frequency–severity modeling with event-informed hyper-parameters

2. Representation learning for high-stakes decisions

I am interested in how auxiliary and unstructured data can be used to learn meaningful latent representations that improve downstream prediction and decision quality, while preserving interpretability and uncertainty awareness.

• Lee, G. Y.  and Okine, N. (Forthcoming on Variance, 2026), Loss reserving with textual description of claims (funded by an individual grant from the CAS).

• Manski, S. and Yang, K. and Lee, G. Y. and Maiti, T. (2021), Loss amount prediction from textual data using a double GLM with shrinkage and selection, European Actuarial Journal. (R and C++ Code, and Data)

• Manski, S. and Yang, K. and Lee, G. Y. and Maiti, T. (2021), Extracting information from textual descriptions for actuarial applications, The Annals of Actuarial Science.

• Lee, G. Y. and Manski, S. and Maiti, T. (2020), Actuarial applications of word embedding models, ASTIN Bulletin.

3. Insurance and risk analytics

A major application area of my work is insurance and actuarial science. I develop models for loss prediction, risk segmentation, ratemaking, reserving, and portfolio-level decision-making, often using modern machine learning tools together with actuarial structure.

• Lee, G. Y. (2025) Long-tail modeling of crop insurance indemnities, Variance.

• Lee, G. Y. and Jeong, H. (2024) Nonparametric intercept regularization for insurance claim frequency regression models, The Annals of Actuarial Science.

• Lee, G. Y. (2021), Regression shrinkage and selection for actuarial models, Variance.

• Lee, G. Y. and Shi, P. (2019), A dependent frequency–severity approach to modeling longitudinal insurance claims, Insurance: Mathematics and Economics.

• Wang, K. and Ding, J. and Lidwell, K. R. and Manski, S. and Lee, G. Y. and Esposito, E. X.  (2019), Treatment level and store level analyses of healthcare data, Risks.

• Lee, G. Y. (2017), General insurance deductible ratemaking, North American Actuarial Journal.

• Frees, E. W. and Lee, G. Y. (2016), Rating endorsements using generalized linear models, Variance (Archive: old articles will be organized here soon).

• Frees, E. W. and Lee, G. Y. and Yang, L. (2016), Multivariate frequency-severity regression models in insurance, Risks.

4. Decision-focused statistical modeling

My research also examines how predictive models interact with practical decisions in finance, insurance, and related areas. The emphasis is not only on model fit, but on decision usefulness, robustness, and real-world constraints.

• Lee, G. Y. (2023) Multivariate insurance portfolio risk retention using the method of multipliers, North American Actuarial Journal.

• Shi, P. and Lee, G. Y. (2022), Copula regression for compound distributions with endogenous covariates with applications in insurance deductible pricing, Journal of the American Statistical Association. (R and C++ Code, and Data)

• Gavagan, J. and Hu, L. and Lee, G. Y. and Liu, H. and Weixel, A. (2021), Optimal reinsurance with model uncertainty and Stackelberg game, Scandinavian Actuarial Journal.

Please use the navigation bar at the top in order to navigate this website!