Hi, I'm Özge Şahin!
I am an assistant professor in the Applied Probability Group of the Delft Institute of Applied Mathematics (DIAM) at TU Delft. I work on statistical machine learning and dependence modeling with copulas, applied to healthcare and finance. I work on patient risk profiling in the RECENTRE consortium and serve on DIAM’s AI in education and AI in research committees.
News
2026
-
I co-organised an AI in education lunch event at DIAM in Delft, the Netherlands.
-
I served as first opponent for Elisabeth Maria Griesbauer’s PhD defence of Vine Copulas in Machine Learning at the University of Oslo, Norway.
-
I presented my work on patient risk profiling with pair-copula constructions at Integreat, University of Oslo, Norway.
-
At the KWG Math & AI Day in Utrecht, the Netherlands, I discussed using AI agents in statistics research and teaching, how I check their outputs, and a course practice tool I built with Claude (slides).
-
I am hosting Umberto Esposito from the University of Bologna, Italy, at TU Delft, the Netherlands, from September to December 2026. We are working on vine copula cluster-weighted models.
News archive
2026
My paper with Matteo Benuzzi and Sandra Paterlini, From hundreds to dozens: stochastic hill climbing for parsimonious environmental, social, and governance scoring, appeared in Annals of Operations Research.
My paper with Thomas Molendijk, Interval-conditioned quantiles of bivariate copulas, appeared in Dependence Modeling.
My chapter with Ozan Evkaya and Ariane Hanebeck, Cluster-Specific Ranking and Variable Importance for Scottish Regional Deprivation via Vine Mixtures, appeared in the Festschrift for Claudia Czado.
I hosted Erasmus+ interns Burak Altuntas and Efe Anik at TU Delft, the Netherlands, from June to September 2026 for work on vine copula modeling in finance.
I gave an invited talk at the Lorentz workshop Mathematics and Machines in Leiden, the Netherlands, on missing data and indicator selection in ESG scoring.
I co-organised DIAM’s AI afternoon in Delft, the Netherlands.
I gave an invited seminar at KU Leuven, Belgium, on vine copula-based classifiers for patient risk profiling.
I joined DIAM’s AI in research committee at TU Delft, the Netherlands.
I built a practice-exam website for Big Data Science in Finance.
2025
My preprint with Annemiek Kwast, Annemieke Witteveen, and Tina Nane, A comparison of variable selection methods and predictive models for postoperative bowel surgery complications, is available on arXiv.
I gave an invited talk at the Eurandom Workshop on Dependence Modeling in Eindhoven, the Netherlands, on conditional quantiles and vine copula-based patient risk profiling.
My preprint Probabilistic patient risk profiling with pair-copula constructions is available on arXiv.
My paper with Ariane Hanebeck, Petra Havlíčková, and Claudia Czado, Sampling from Conditional Distributions of Simplified Vines, appeared in Statistics and Computing.
I gave an invited talk at the Dutch Math Finance Afternoons in Utrecht, the Netherlands, on missing data and indicator selection in ESG scoring.
I hosted Gökçe Dayanıklı from the University of Illinois Urbana-Champaign, United States, for a three-day research visit to TU Delft, the Netherlands, on dependence modeling and game theory.
I co-organised and co-taught the tutorial From Theory to Practice: Vine Copula Models at DAGStat in Berlin, Germany.
My work received the Journal of Classification’s “Highlighted paper of the year” recognition.
I began co-supervising Umberto Esposito at the University of Bologna, Italy, with Gabriele Soffritti, on vine copula cluster-weighted models.
2024
I gave an invited talk at the Finance Research Day at TU Delft, the Netherlands, on pitfalls in ESG scoring.
My paper with Harry Joe, Vine Copula-Based Classifiers with Applications, appeared online in Journal of Classification.
I hosted Ariane Hanebeck from the Technical University of Munich, Germany, for a three-day research visit to TU Delft, the Netherlands, continuing our work on sampling from conditional distributions of simplified vines.
I co-organised the Workshop on Dependence Models, Vines, and Their Applications in Munich, Germany.
I hosted Claudia Czado from the Technical University of Munich, Germany, for a three-day research visit to TU Delft, the Netherlands, to continue our collaboration on vine copulas.
My paper with Claudia Czado, High-dimensional sparse vine copula regression with application to genomic prediction, appeared in Biometrics.
I gave an invited seminar at the University of Edinburgh, United Kingdom.
I began the daily supervision of Victor Ryan’s PhD at TU Delft, the Netherlands, on distribution shifts in copula-based modeling with healthcare applications, with Tina Nane and Annemieke Witteveen, within RECENTRE.
I joined DIAM’s AI in education committee at TU Delft, the Netherlands.
2023
I presented at the 16th International Conference of the ERCIM WG on Computational and Methodological Statistics in Berlin, Germany.
I co-organised and co-taught the tutorial Risk management with vine copula based dependence models in Berlin, Germany.
I gave a talk at the 15th DIAM Research Day at TU Delft, the Netherlands.
I joined TU Delft, the Netherlands, as an assistant professor in the Applied Probability Group.