Modeling citizens’ behavioural patterns towards housing and mobility mode for algorithmic urban policy design

International audienceAs cities become increasingly populated, urban design and policies play a key role in ensuring the resilient, sustainable, safe, and inclusive development of metropolitan areas. CityScope is a data-driven tangible platform that aims at bringing different stakeholders together to collaboratively decide which are the most adequate urban actions to be taken. The algorithmic zoning project described in this document constitutes a new module of CityScope. The idea behind this module lies on the exploration of dynamically reconfigurable set of urban policies that will contribute to an open, cross-sectoral, and informed decision making, while promoting greener, more walkable, and diverse urban areas. This abstract summarizes the first step towards the aforementioned dynamic urban policy system based on incentives. Among the numerous aspects that the policy system could address, we will focus on the deficiency of affordable housing in city centres. In order to foresee citizens’ reactions to a wide range of housing-related financial stimuli, it is necessary to unveil the dynamics that take place among the agents constituting the system. This includes characterizing citizens’ criteria when choosing their housing option –and, thus, their residential mobility mode–. CityScope will then be able to interactively suggest those policies based on incentives that boost the urban indicators that best reflect citizens’ aspirations regarding the built environment. The present methodology suggests the development of an Agent-Based Model where housing- and mobility-mode-related preferences are described based on eight different income profiles. The housing-related criteria include (1) housing price, (2) diversity acceptance, (3) zone preference, and (4) commuting time. As this last parameter depends on the availability of transportation and users’ preferences, this joint model considers the interconnection between housing and residential mobility. Apart from the (1) resulting commuting time, mobility-related criteria consist of (2) commuting price, (3) social pattern, and (4) difficulty of usage. The simulation process is inspired by the well-known Schelling segregation model. An iterative process has been designed, so that in iteration i+1 citizen agents are given an alternative housing option to the one they had chosen in iteration i. After evaluating the most convenient available transportation mode for this alternative housing option, and the resulting commuting time, they will decide whether this option maximizes their preference score. The simulation converges whenever the amount of agents willing to move to an alternative housing option asymptotically approaches zero. In order to represent the influence of external regions surrounding our area of interest, while working with a reasonable amount of environmental variables, two levels of granularity have been defined: (1) the census block group level and (2) the building level. The geographic data used for the simulation comprises Kendall Square and the Greater Boston Area, in Massachusetts, US. The criteria have also been calibrated for this particular zone, using census data and transportation data. The identification and tuning of these parameters, therefore, enables the prediction of citizens’ reactions to urban disruptions and potential financial incentives tested on the CityScope platform. The algorithmic proposal of the most beneficial actions is, then, possible