This postdoctoral project investigates how explainable artificial intelligence can support the analysis of spatial and metabolic differences among Belgian cities. Its purpose is not simply to classify cities, but to make the variables driving each grouping understandable to researchers, planners and public authorities.

The research assembles spatial, environmental, infrastructural and resource-use indicators and uses data-driven clustering to identify recurring metabolic profiles. Explainability techniques are then used to determine which features distinguish the clusters and how these relationships vary across space. This creates a bridge between computational pattern recognition and the substantive interpretation required for urban policy.

The project contributes to the Lab’s work on metabolic geographies by testing how advanced analytical methods can remain transparent and decision-relevant. It also complements the relational perspective of ReFrameCity: while one strand examines interurban connections conceptually, the explainable-AI strand identifies empirical similarities and differences within the Belgian urban system. The work is supported through Wallonia-Brussels International cooperation.