Paramétrage d’un modèle de simulation individu-centré des mobilités résidentielles : enjeux, démarche et résultats

Abstract

The parameters of an agent-based model, in social sciences, reproduce human behavior, and their quantification are often based on empirical data. Agent-based models are commonly used and described in the literature but the calibration protocol process is less documented. To broaden these ongoing discussions, this article presents an approach for calibrating an individual-based residential simulation model. The model is integrated in an agents-based LUTI simulation platform - Mobisim – which aims at simulating the impact of spatial change scenarios on daily and residential mobility. The simulation outputs can then be used to support land planning decision making. The model calibration has been designed according to the model formalization choice and to the objectives it has been developed for. The adopted procedure is iterative and empirical. It relies on a combination of different methods, related to the availability and quality. The relevance of the chosen values to estimate the parameters and the outputs consistency has been verified through exploratory analysis.

Keywords

• simulation
• residential mobility
• agent-based model
• mobility behavior
• modeling/modelling