شبیه سازی مبتنی بر عامل رفتار مصرف کننده در خرید خواروبار در سطح منطقه ای

In the field of multi agent systems, spatial modeling is gaining new momentum. Spatial modeling offers the possibility of simulating human actions on a micro scale level in a way that blends in with the existing network of theoretical approaches in the social sciences. The present research project examines the applicability of techniques that simulate spatial choice in shopping behavior on a regional level. This paper presents the particulars of an agent-based micro model for grocery shopping, based on an individual population and store data gathered in northern Sweden. The high quality of the data, gathered exclusively for this project, allows a fine validation of the simulation results. Future applications include the prognosis of consumer behavior and turnover forecasts on the basis of which alterations could be made on the supply side, particularly regarding the competition among stores in the center and those on the edges of cities.

With regard to the role of geography in the retail trade, consumer choice behavior reflects the important but often ignored demand side of the economic process of shopping. Spatial consumer behavior and the spatial structure of retailing are dynamically interrelated. Although the consumers, by their choice of a particular shopping location, form the economic base of the retailer, the latter may influence the consumers by adopting different business strategies. Agergård et al. (1970) were the first to analyze these phenomena, describing the development of retail locations in urban centers as a spiral movement. A current stage in this development is the growing competition between retail locations in the center and those on the edges of the cities, which is increasingly perceived as a threat to the urban business environment itself. Planning is therefore concerned with finding more effective research methods to judge the potential impacts of such developments. At the same time consumer attitudes and behavior are changing rapidly toward more individuality and more diversity. Agent-based simulations are currently the most promising tool to address these challenges on the level of modeling.

Consumer behavior is embedded in a triangle consisting of the consumers themselves, the structure of the retail environment and the institutional context of planning and politics, other service infrastructure (e.g. transportation), and the social context, including behavioral rules, social standards and so forth. This triangle is subject to change in time and space, which means that the area of simulation practices is extensive. On the demand side, fine-scale population prognoses may be the basis for exploring future developments of buying power flows under circumstances of further (sub)urbanization. On the supply side, an area of research may be the impact of already observable or expected changes in the spatial structure of retailing on consumer movements and spending power allocation. This could open a new perspective on the discussions about the competition between city center and town edge retail locations. Finally, although Arentze and Timmermans (2005) as well as O'Sullivan and Haklay (2000: page 1409) criticize agent-based modeling for neglecting the institutional contexts of the simulated individual decisions in the past, this approach may in fact be better capable of incorporating such phenomena than other methods are (Koch, 2003 and Koch, 2005). A further outstanding advantage of multi agent simulations in comparison to other individual-based modeling approaches is the possibility to integrate communications among individuals. While little literature is available on how such communications affect spatial consumer behavior, initial randomly generated small world networks may constitute a useful paradigm for studying the mechanisms of social interactions in the process of choosing shopping locations.