Postdoctoral Fellow
Research group|Statistics and Data Science
Main supervisor|?Thordis Linda Thorarinsdottir
Affiliation |?Department of Mathematics, UiO
Contact |?touqeera@math.uio.no
Short bio
I recently joined the Department of Mathematics at UiO as a postdoctoral research fellow under the DSTrain–MSCA program. The project will focus on the stochastic modelling of natural processes. Before that, I was a postdoctoral research fellow at the Institut Denis Poisson, Université d’Orléans, France, from May 2025 to October 2025. There, I worked on the missing data problem in high-dimensional environmental time series as part of the JUNON project. Prior to that, I also held a postdoctoral position at ENSAI, France, from May 2023 to April 2025, where I worked on machine learning and data mining of extreme values in massive data under a prestigious Région Bretagne research grant. I earned a PhD in Statistics with the Doctor Europaeus Label in June 2023 from the Department of Statistical Sciences at the University of Padova, Italy. My doctoral research focused on the modeling of discrete extreme values. During my PhD, I also worked as a visiting research scholar at the Laboratoire de Mathématiques de Versailles, Université de Versailles Saint-Quentin-en-Yvelines, and LSCE, Paris, France, as well as at the Research Center for Statistics, University of Geneva, Switzerland. For more information, please visit my personal webpage https://touqeerahmadunipd.github.io/
Research interests and hobbies
Drawing from my research focus on machine learning and advanced time series models, I intend to incorporate adaptable approaches that boost inference and prediction in the stochastic modelling of natural phenomena. Specifically, my DSTrain project will concentrate on the enhancement of modelling for drought extremes. The nonstationary of spacetime will be addressed by the implementation of a distributional regression framework. Furthermore, the investigation encompasses neural network-based distributional regression models utilizing deep Bayesian neural networks for one-step-ahead extreme event prediction. I am interested in hybrid Neural-GAM models that improve interpretability in time series analysis, spatio-temporal Bayesian deep learning for environmental forecasting, and functional data analysis employing deep learning for continuous time series extremes. My also intended to examine nonparametric machine learning methodologies, causal inference in time series and dimensionality reduction contexts, and deep generative models to tackle issues such as absent data and irregular structures in environmental datasets. Furthermore, I intend to further my research on uncertainty quantification in time series compounding extremes by integrating bootstrap-based confidence intervals with Bayesian techniques to deliver robust prediction intervals for high-dimensional environmental applications. These guidelines will facilitate the creation of more dependable, interpretable, and adaptable models for forecasting inside intricate data frameworks.
I enjoy playing badminton and cricket, trekking and hiking, and cooking, which let me express creativity beyond research. I also love traveling, as it allows me to explore new cultures and perspectives.
DSTrain project
Flexible modelling of extreme natural processes (SMNP)
The DSTrain project aim to advance the drought characterization and risk assessment by developing innovative statistical modeling frameworks that transcend traditional stationary approaches. The core objective is the creation of a flexible, nonstationary spatio-temporal modeling framework, which integrates heavy-tailed distributions, zero-inflation, and compound extreme event analysis. This methodology enhances conventional drought metrics into more adaptive Extended Drought Indices.
The primary purpose of this research is the development of the extended drought indices and their nonstationary counterparts. These indices provide a more accurate quantification of drought severity, frequency, and duration by capturing both upper and lower precipitation extremes. The analytical framework employs distributional regression to model nonlinear relationships between drought behavior and key climate covariates, including temperature, evapotranspiration, and soil moisture. A critical innovation is the explicit incorporation of compound extreme event analysis, such as the co-occurrence of droughts and heatwaves, to assess amplified environmental and agricultural impacts.
The models will be rigorously validated through simulation experiments and applied to historical climate data from European catchments. These applications yield insights into centennial-scale drought variability and future drought hazards. The project deliverables include refined drought indices, predictive algorithms for extreme events, and open-source computational tools in R and Python. By bridging theoretical climate modeling with practical risk assessment, this research provides robust tools for enhancing water resource management, agricultural resilience, and disaster mitigation, thereby strengthening socio-ecological systems against a changing climate.
?
Publications
DSTrain publications
For all publications, please visit my google scholar page.
https://scholar.google.com/citations?user=0Unv8IAAAAAJ&hl=en
Previous publications
?
?