Lead supervisor:
Dr Hamidreza Eskandari 
Other supervisor(s):
- Dr Daniele Doneddu
Start date: October 2025 (Full time)
Award: Collaborative
Subject Pathway:
Management and Business
Thematic Cluster:
Economy, Enterprise, and Productivity Cluster
In partnership with:
Predicting and Mitigating Energy Poverty in Wales: An Explainable Machine Learning Approach for Early Intervention
This research establishes a framework for early-warning risk analysis by applying spatio-temporal predictive modeling to investigate how systemic macroeconomic shocks translate into localized household vulnerability over time. By linking multi-horizon price dynamics across essential commodities with longitudinal household surveys, the analytical architecture decomposes systemic hazards into micro-level exposure and absorptive capacities.
The modeling evaluates hardship risk across sequential time horizons while accounting for geographical variations and regional infrastructure constraints. Forward-looking hazard trajectories are integrated with household-level socio-economic profiles through dynamic classification and survival modeling techniques. This allows the framework to assess both the spatial dispersion of vulnerability and the temporal pace of shock transmission, establishing rigorous empirical bounds on how effectively forward-looking signals can identify emerging household distress before it occurs.
Research Impact
-
Reframing Vulnerability and Hazard Screening: Demonstrates how macro-level forecasting models interact with micro-level lived realities, illustrating that aggregate market indicators must be integrated with local and dwelling-level exposure data to reliably forecast household vulnerability.
-
Overcoming Spatial and Structural Blind Spots: Exposes critical spatial misalignments between conventional income-based poverty metrics and real-world vulnerability, providing policy bodies with the evidence base needed to reform support schemes that overlook regional infrastructure dependencies.
-
Translational Decision-Analytic Policy Tools: Connects predictive statistical modeling to real-world deployment through decision-curve evaluation, identifying which administrative markers deliver the highest operational benefit for targeted intervention during rapid economic crises.
-
Advancing Spatio-Temporal Risk Methodology: Sets a transferable standard for socio-economic early warning systems by pairing out-of-time temporal validation with spatial modeling, providing a scalable template for resilience planning, social protection design, and crisis auditing.
Bibliography
-
Boardman, B. (1991). Fuel Poverty: From Cold Homes to Affordable Warmth. London: Belhaven Press.
-
Hyndman, R. J., & Koehler, A. B. (2006). Another look at measures of forecast accuracy. International Journal of Forecasting, 22(4), 679–688.
-
Middlemiss, L., & Gillard, R. (2015). Fuel poverty from the bottom up: Characterising household vulnerability through the lived experience. Energy Policy, 82, 146–154.
-
Nosek, B. A., Ebersole, C. R., DeHaven, A. C., & Mellor, D. T. (2018). The preregistration revolution. Proceedings of the National Academy of Sciences, 115(11), 2600–2606.
-
Robinson, C. (2019). Spatialising fuel poverty: A geographical perspective on vulnerability and governance. Local Environment, 24(7), 629–647.
-
Vickers, A. J., & Elkin, E. B. (2006). Decision curve analysis: A novel method for evaluating prediction models. Medical Decision Making, 26(6), 565–574.
Biography
Mohsen Asghari Ilani is a data scientist and researcher specialising in machine learning applications for socio-economic, energy, and environmental risk analysis. He is currently a doctoral researcher at Swansea University funded by the ESRC/UKRI through the Wales Graduate School for the Social Sciences (WGSSS). His research focuses on predictive spatio-temporal modeling, energy transition dynamics, and early-warning frameworks for household vulnerability, alongside broader work in environmental management and deep learning.
Mae Mohsen Asghari Ilani yn wyddonydd data ac yn ymchwilydd sy'n arbenigo mewn cymwysiadau dysgu peirianyddol ar gyfer risgiau cymdeithasol-economaidd, ynni ac amgylcheddol. Ar hyn o bryd, mae'n ymchwilydd doethurol ym Mhrifysgol Abertawe a ariennir gan ESRC/UKRI drwy Ysgol Graddedigion y Gwyddorau Cymdeithasol Cymru (WGSSS). Mae ei ymchwil yn canolbwyntio ar fodelu gofodol-amserol rhagfynegol, deinameg trawsnewid ynni, a fframweithiau rhybudd cynnar ar gyfer bregusrwydd aelwydydd, ochr yn ochr â gwaith ehangach mewn rheolaeth amgylcheddol a dysgu dwfn.