Collection Mémoires et thèses électroniques
Accueil À propos Nous joindre

CHAPTER 10: Use the Shopping Behavior Multiagent Geosimulation as a Spatial Decision-Making Tool

Table des matières

In this chapter, we present the description of the last step of our method. This step shows how end-users can use multiagent geosimulations to support spatial decision-making. It also illustrates this step using the shopping behavior case study. Therefore, it demonstrates how mall managers may use the shopping behavior multiagent geosimulation prototype to make some decisions about their mall configuration.

Simulation is generally used to analyze what-if scenarios (or alternatives). It can be used to study and compare alternative designs, or troubleshoot existing systems (Simon, 1960). This allows users to determine the impact of changing key variables in simulation models and processes. Using simulation models, we are free to imagine how an existing system might perform if altered, or imagine, and explicitly visualize, how a new system might behave before the prototype is even completed. The ability to easily construct and execute models, as well as generate statistics and animations about results, has always been one of the main attractions of simulation applications. The nature of simulation makes it a good tool to support decision-making. Furthermore, if we deal with the spatial decision-making process, it is appropriate to use simulation that deals with spatial and geographic data, such as geosimulation. In the literature, the decision making processes that include spatial information are known as spatial or geographic decision-making processes . They are generally more complex than the classical decision-making processes, which do not involve spatial data (Densham, 1991). This kind of spatial decision is also called Spatial Decision Problem (SDP) , which is characterised by the following features (Simon, 1960):

- A large number of decision alternatives;

- The outcomes or consequences of the decision alternatives are spatial variables;

- Each alternative is evaluated on the basis of multiple criteria;

- Some of the criteria may be qualitative, while others may be quantitative;

- There is typically more than one decision-maker involved in the decision-making process;

- The decisions are often surrounded by uncertainty.

In this chapter, we present the last step of our proposed method. This step concerns the use of the geosimulation by end-users. It presents how geosimulation can help them to make decisions related to a phenomenon’s behavior or to assess the configuration of a geographic environment. This chapter also presents the illustration of this step using the shopping behavior case study.

This chapter is organized as follows: Section 10.1 presents a generic description of the step. In Section 10.2, we present the illustration of the step using the shopping behavior case study. Therefore, we show how mall managers can use the shopping behavior geosimulation prototype to make decisions about spatial configurations in their malls. Section 10.3 then discusses the chapter in relation to other works. Finally, Section 10.4 concludes the chapter.

Before presenting the details of the step, let us define some basic concepts:

- Configuration: It is defined in the Oxford English Dictionary as follows: ‘ Arrangement of parts or elements in a particular form or figure; the form, shape, figure, resulting from such arrangement; conformation; outline, contour (of geographical features, etc.) .’ (

- Environment configuration: It is the spatial arrangement of the elements belonging to this environment (their geographic features, shapes, positions, etc.).

During this step, a user can use the geosimulation in order to explore various spatial configurations of a simulated environment. For each simulation scenario involving new configuration, of the environment, the user can launch the geosimulation, collect the generated outputs, and then analyze them. By comparing these results, he can make informed decisions about the impact of spatial changes in the environment. This step contains the following sub-steps (see Fig 10.1):

(1) Prepare some simulation scenarios : In this sub-step, a user prepares some simulation scenarios. In each scenario, the user can change the spatial configuration of the simulation environment. For example, he/she can change the position of an entity belonging to this environment, the spatial configuration of this entity, etc.

(2) Launch the geosimulation application using the scenarios : In this sub-step, a user can launch the simulation for each scenario.

(3) Generate simulation outputs and analyze data for each scenario : In this sub-step, a user generates and analyzes data gathered from each simulation scenario, and stores the output data in files or databases related to each scenario.

(4) Compare the simulation scenarios : In this sub-step, a user compares the results of the analyses of the scenarios. Based upon this comparison, a user can decide about the best scenario and may apply it in the real environment.

This step is illustrated in the next section (Section 10.2) using the customers’ shopping behavior in a mall.

This section aims to illustrate the last step of our method using the shopping behavior case study. In this illustration, we demonstrate how mall managers use of Mall_MAGS prototype for the spatial decision-making process. Before presenting this illustration, it is relevant to present, in detail, why these managers need this geosimulation tool of shopping behavior in a mall.

The world of shopping malls has been changing dramatically in the last decade, buffeted by, among other things, the introduction of electronic commerce, the saturation of locations, and changes in customers’ shopping behavior (Ruiz et al., 2004). According to (Wakelfield et al., 1998), there are essentially three factors which explain the mall’s declining role. First, consumers are increasingly busy, have less time for shopping, and therefore, reduce the frequency of their visits to the mall. Next, too many malls are alike, and customers will go to the shopping center that offers the most product and service variety, as well as the most comfortable atmosphere. Finally, (Wakelfield et al., 1998) emphasize the fact that fewer consumers are going to the mall in order to «enjoy their shopping experience». These factors lead mall managers to develop strategies to differentiate their malls from the competition, in order to enhance customer loyalty.

In a mall, the interaction between people and the environment is an important issue (Norman 1988). The environment can be characterized by its degree of complexity, mystery, coherence, and legibility (Kaplan and Kaplan 1989). Perception plays a key role in customers’ activities in a mall (Sheridan, 2002). Spatial legibility may be thought of as a way for mall managers and tenants to communicate information to customers. Hence, the importance of stores’ locations, the perception of products, or shopping opportunities, when customers walk through the mall’s corridors. Public buildings that are not legible often induce frustration and negative reactions on busy people who cannot easily find their way. In shopping malls, spatial legibility is of great importance, and one of the important related issues is the mall layout (Garling et al. 1982). Indeed, people have different methods of finding their ways in complex spatial environments such as memorizing particular locations and routes in information-rich environments, like malls. One important property of a legible space is to facilitate the creation of mental maps of its layout by the individuals who frequent this space. Hence, the importance of adequately locating stores in a mall (Hernandez and Biasiotto, 2001). During a shopping trip in a mall, a customer may have a precise idea of where he or she wants to purchase given items. However, discovering unexpected shopping opportunities also influences a customer’s decision–making, and may result in a buying decision if the opportunity fits with the customer’s needs and preferences. Creating buying opportunities for a large proportion of customers is an important goal for mall managers and tenants. To this end, mall managers must find a location for every store that will optimize the chances for customers to be attracted to this store and by the buying opportunities it may offer. Managers know very well the importance of ‘ anchor stores ’, such as Wall Mart which attract certain types of customers and favor ‘ proximity shopping ’ in stores that are located in corridors converging toward the anchor stores (Konishi and Standfort, 2001).

Changing a mall configuration is a very important and expensive decision in terms of money and time. In order to guarantee the success of such a decision, mall managers should be able to better understand customers’ behaviors, and the way they may react to the changes in the mall’s configuration. Certain traditional techniques may help mall managers to understand how customers interact with the mall environment. For example, they can use questionnaires to collect information about customers and analyze the collected data in order to try to understand how customers use the mall. Although surveys can help mall managers to understand how customers appreciate the current mall configuration (and it is well known that most customers are not keen on filling out questionnaires), they are not very useful for anticipating the reactions of customers to future changes in the mall’s configuration. Hence, managers lack tools to anticipate customers’ reactions to changes in the mall’s configuration.

Indeed, optimizing the location of stores in a mall is a complex problem if a manager wants to take into account the factors that influence the customers’ shopping and buying decisions in relation to the mall’s spatial layout: 1) relative locations of stores; 2) store’s location in relation to corridors, entrances, and other services; 3) customers’ preferences in relation to their needs and socio-economic profiles; 4) Customers’ perception of buying opportunities in the mall in an environment that is rapidly changing. Traditional statistical and data analysis methods are not able to take into account so many factors, and cannot encompass the spatial and perceptual characteristics of people’s shopping behaviors.

An ideal solution would be to enable mall managers to try various mall configurations by changing the locations of certain stores and to carry out surveys in order to determine the impact of these changes on customers. Obviously, such a solution is not practical in a real setting because: 1) changing a store’s location is a costly activity that cannot be done often; 2) it is not possible to try several locations for a store and to assess the reactions of customers for each of these locations before making a final decision about the store’s location. An alternative solution would be to simulate on a computer the customers’ behaviors in a virtual mall and enable managers to explore various scenarios by changing stores’ locations in the virtual mall and by observing the reactions of customers to these changes. Until recently, such an approach was not feasible; however, thanks to recent progress in the areas of geosimulation (Benenson and Torrens 2004) and multiagent systems simulation (Moss and Davidson, 2001) and more specifically, multi-agent geo-simulations (Moulin et al. 2003), simulating the behaviors of a large number of virtual agents in a georeferenced virtual world is now possible.

To illustrate the use of the shopping behavior geosimulation tool we used two simulation scenarios. In the first, we launched a simulation with a given configuration of the shopping mall (Fig 10.2), and with a population of 390 shoppers. The simulation generated output data regarding shoppers’ itineraries during their shopping trips. In scenario 2, we switched the locations of a two department stores: Wal-Mart and Zellers (Fig 10.3). The simulation was once again launched, and generated the output data for the shopper agents’ itineraries once again. By comparing the output data from the two scenarios, we noticed the divergence between the paths taken by the shopper agents in order to reach the department stores. The simulation analysis showed that corridor X was less frequented in scenario 2 than it was in scenario 1 (Fig 10.2). However, corridor Y was more often used in scenario 2 than in scenario 1 (Fig 10.3). In these figures, the flow of the shoppers agents passing through a corridor is represented by a line, attached to the corridor. The width and color of the line are proportional to the flow of shopper agents that use the walkway. If the flow increases, the width of the line grows, in addition to its color becoming darker. Through a data analysis of the shopper agents’ characteristics, we can see that in scenario 2, most of the agents going through corridor Y are female, and they go to the mall primarily to visit female clothing stores. If the mall manager chooses the mall’s configuration portrayed by scenario 2, he may consider renting the spaces along corridor Y to female clothing stores.

Determining store locations in a mall is widely recognized as the most decisive factor in defining retail success or failure. As it has been observed over many years, good locations are ‘the keystone to profitability’ (Hernandez and Biasiotto, 2001). They represent an aspect of major investment that must be managed. Once made, poor location decisions are difficult to remedy, and it is these factors that, in theory, ‘compel the retailer to make the decision carefully’ (Hernandez and Biasiotto, 2001). Due to increasing competition, the pressure placed upon retailers to make ‘good’ decisions has grown markedly, as the consequences of ‘bad’ decisions have escalated. For these reasons, retailers, such as mall managers or store managers, who need to make decisions about their locations, need to be supported in these decisions by efficient tools.

The main objective of a simulation tool is to aid users in their decision-making process with regards to a phenomenon, or to help solve problems related to this phenomenon. When we deal with a spatial phenomenon (i.e., which behaves in spatial environment) the decision-making process is generally more complex than that of the classical decision-making process, given that it involves spatial data (Densham, 1991). If we look at simulation applications proposed in the literature to simulate human behaviors in spatial environments (Raubal, 2001), (Dijkstra, 2001), (Koch, 2000), and (Moulin et al., 2003), we note that they are used only to visualize the simulation course. Thus, they are not used to compare simulation scenarios, nor to generate results about the comparisons of these scenarios. This represents a noteworthy gap in the usage of simulation. In order to fill this gap, we developed a multiagent geosimulator of shopping behavior that can be used to compare several simulation scenarios. This prototype generates analysis results related to these comparisons, and can be applied by users (mall managers) in order to make efficient decisions about the phenomenon to be simulated, or about the simulation environment.

To sum up, our shopping behaviour geosimulation tool stands out from other human behavior simulation tools because it can be used by end-users to make spatial decisions about the configurations of the simulated environment. This can be considered as a good contribution in the field of simulation.

This chapter presented and illustrated the last step of our method which is the use of the geosimulation by its end-users. First, we presented why it is appropriate to use geosimulation tools in order to aid users in their decision-making process. We focuzed on the importance of using the geosimulation as a spatial decision-making tool for spatialized behaviors in geographic environments. Then, we illustrated the step using the customers’ shopping behavior in a mall as a case study. Hence, we showed why and how users, mainly mall managers, can use our shopping behavior geosimulation in order to try various mall configurations, by changing the locations of certain stores. As mentioned in this chapter, we can see that the shopping behavior prototype can satisfy the main users’ needs set in the first step of our method (Chapter 4). The users (mall managers) of the shopping behavior prototype, can visualize the flows of shopper agents in several simulation scenarios. Using our analysis tool and the OLAP/SOLAP one they can visualize in a graphical form the characteristics of these shopper agents and combine them and compute and compare several indicators of interest. What’s more, mall managers can exchange the positions of certain stores and observe how virtual shoppers react to these configuration changes, etc. They can again use the analysis and OLAP/SOLAP tools to asses the different scenarios and compare them. We think that this prototype can help mall managers to make better decisions about the mall configuration (store locations), as well as the impact of changes on customers’ shopping behavior. Let us mention that we presented the simulation prototype to mall managers of another mall, Place De La Cité in Quebec City, and they showed a great interest in using our simulation tool. They asked us to develop, as soon as possible, a simulation prototype for their mall, using the data collected during our 2004 survey about shoppers attending Place De La Cité mall.

© Walid Ali, 2006