From Functional Planning to Hybrid Approaches: Rethinking the Modelling of Complex Urban Dynamics

Urban modelling has evolved since the 1950s, moving from classic, linear and deterministic models to systemic approaches capable of better understanding the complexity of territorial dynamics. While the first models helped to organise urban planning, they proved to be limited in the face of the unpredictability of social, economic and environmental interactions. The introduction of complexity theory paved the way for computational and multi-agent approaches, which are better suited to emerging phenomena. In this context, structural equation modelling (PLS-SEM) has emerged as a key methodological tool, allowing latent variables to be integrated and complex causal relationships to be modelled. More recently, artificial intelligence, particularly Random Forest algorithms and their hybrid combinations with PLS-SEM, has enhanced the explanatory and predictive power of these models. This review offers a chronological and critical overview of these developments, highlighting the contributions, limitations and prospects of current methods.