Collection Mémoires et thèses électroniques
Accueil À propos Nous joindre

CHAPTER 6: Collect and Analyze Data Used as Input for the Multiagent Geosimulation

Table des matières

In this chapter, we present the step in which we collect and analyze the simulation input data. This data is used to feed the agent-based simulation models, which were presented in the previous chapter. It also illustrates this step using the customers’ shopping behavior in a mall as a case study.

In the previous chapter, we presented the step in which we design the multiagent geosimulation models of the phenomenon to be simulated and its environment. We also illustrated this step by designing the shopping behavior multiagent geosimulations models. These models are based upon the agent technology and take into account the spatial features of the agents composing them. In order to execute the simulation models on a computer, we need data. Since we deal with geosimuation, a significant part of this data needs to be geographic or spatial.

This chapter aims to present the fourth step of our method, which aims to gather and analyze input data used to feed the multiagent geosimulation models. It also illustrates this step using the shopping behavior as a case study. The remaining part of the chapter is organized as follows: In Section 6.1 we present a generic description of the step. Section 6.2 briefly presents our case study, which is the Square One shopping mall in Toronto, Canada. In Section 6.3, we present how we gathered and digitalized our simulation input data concerning shopping behavior in a mall. Next, Section 6.4 aims at presenting the analysis results of the gathered data. Then, in Section 6.5, we discuss the material presented in this chapter, which is followed by Section 6.6, the chapter’s conclusion.

This section aims to present generic descriptions of the fourth step of our proposed method. This step aims to gather the data which is used to feed the multiagent geosimulation model designed in the previous step. It also aims at analyzing this data.

This step has as inputs the initial version of the model of the phenomenon to be simulated (designed in the second step) and the multiagent geosimulation model designed in the previous step (third step). It contains two main sub-steps which are presented in the following points (see Fig 6.2):

(1) Collect the data : In this sub-step, we collect data and transform it in order to feed the simulation models. If we provide the simulation models with random data, this sub-step can be very simple. However, if we want to use real data, we must collect it before feeding it in the system. Several techniques can be used to collect non-spatial and spatial data. Here are some techniques: questionnaires/surveys on paper, digital questionnaires, Web-based questionnaires, observations, consulting experts, GPS (Global Positioning Systems) for spatial data, etc. In this sub-step we must:

- choose the simulation case test: Before collecting data, we must choose the case test which is used in the simulation;

- collect data about the phenomenon to be simulated: Here, we must collect non-spatial and spatial data concerning the phenomenon to be simulated. This data is used to feed the non-spatial and spatial structure of the active agents representing the system to be simulated (see the multiagent simulation models developed in the previous steps).

- collect data about the simulation environment: Here, we must collect geographic and non-geographic data concerning the simulation environment. This data is used to feed the non-spatial and spatial structure of the passive agents representing the elements of the simulation environment (see the multiagent simulation models developed in the previous steps). Generally, data about the geography of environments is stored in geographic information systems (GIS).

If we decide to use surveys to collect data, we propose the following directives to be followed (see Fig 6.1):

■ Preparing the survey’s structure and content: The preparation of the survey’s structure is based upon the initial version of the model of the phenomenon to be simulated and the multiagent model of the geosimulation. Each factor or variable belonging to these models is formulated by one or more questions in the survey. For each question, we must decide about its formulation and the answers’ alternatives. We also need to think about the structure of these answers (one answer, multiple answers, etc.). Once the survey is built, it must be pre-tested before administrating it to the target audience. Hence, we can choose a certain number of persons (colleagues, friends, etc.) and ask them to answer the questions of the survey. After the pre-test, we can modify the survey by adjusting its structure or content. After several adjustments, the survey is ready to be conducted with the target audience.

■ Conducting the survey: Surveys will be conducted by qualified persons. These persons must be hired and trained to conduct the survey. We also must choose the time and place where the survey will take place. Sometimes, if the survey is intended to collect data about human beings, we need to respect some ethic clauses or constraints before conducting the survey.

■ Digitalizing the survey data: In order to develop operational computer geosimulations, we need digital data. If the collected data is already digital, we do not need the digitalization process; otherwise (if we use a paper survey) we must digitalize it. To digitalize the data, we must design the structure of files or database intended to contain the data and enter the data in these files or database using an appropriate specific software.

(2) Analyze the data : Once the data are gathered and digitalized, it is relevant to analyze it in order to determine some behavior patterns or categories of the phenomenon to be simulated. Since we deal with both non-spatial and spatial data, we propose to analyze these two kinds of data.

- Non-spatial analysis: In this category, we analyze the gathered data related to non-spatial variables. These variables exist in the initial version of the model as non-spatial factors. They exist in the multiagent geosimulation model as attributes of the agents (passive or active). The non-spatial analysis results inform us about the distribution of the data based upon the non-spatial variables. Based upon these results, we can define some non-spatial behavior patterns or categories of the phenomenon to be simulated. If the analysis involves one variable, it is called uni-variable non-spatial analysis; otherwise, it is called multi-variables non-spatial analysis .

- Spatial analysis: In this category, we analyze the gathered data related to spatial variables. These variables represent the spatial characteristics of the simulation environment in both the initial version of the model and the multiagent geosimulation one. The spatial analysis gives us ideas about the frequence of use of the spatial elements belonging to the environment. If the analysis involves one variable, it is called uni-variable spatial analysis , otherwise it is called multi-variables spatial analysis .

- Spatial and non-spatial analysis: In this category we analyze the combination of non-spatial and spatial variables. This analysis gives us ideas about the use of spatial elements belonging to the environment by other non-spatial elements existing in the simulation. Based upon the results of such analyses, we can define some spatial behavior patterns or categories of the phenomenon to be simulated.

In the remaining part of the chapter we present the illustration of this step using the customers’ shopping behavior as a case study.

This section aims to present our geosimulation test case, which is the Square One mall. Owned by OMERS Realty Corporation and operated by OMER Realty Management Corporation (http://www.omers.com/scripts/index.asp), the Square One shopping mall (http://www.shopsquareone.com/) is located in Mississauga, Ontario. It is Canada’s second largest mall in terms of the number of stores and services offered. With approximately 350 stores and services, it is visited by twenty millions pedestrians each year, averaging 350,000 per week. Square One is the largest shopping mall within the greater Toronto area (GTA), which is the fifth largest metropolis in North America. Mall tenants include Canada’s top chains, America’s best retailers, and unique speciality shops (OMERS, 2001). Moreover, four major department stores: The Bay, Sears, Wal-Mart, and Zellers anchor the mall, with the latter two being the largest of their kind in Canada. The mall consists of two levels. The upper level boasts a variety of mid to high-end stores with an emphasis on the latest fashions and accessories. The lower level continues the fashion emphasis, but to a more youthful market, and includes speciality tenants. The mall has ample parking and access via local transit, which brings approximately 4.5 million customers annually, accounting for 23% of total visits (shoppers visiting the mall). The mall is well positioned in terms of its trade area, with one of Canada’s highest household income values at $67,000 per year. The city of Mississauga grew by 81,000 people between 1991 and 1996 - the largest population increase in any Canadian city. Sales productivity levels for the mall are some of the highest in Canada, at $585 per square foot (OMERS, 2001).

Our shopping behavior simulation prototype will be used by shopping mall managers in order to make decisions. For this reason, the simulation prototype needs to be close to reality, or ‘realistic’. In order to have a ‘realistic’ shopping behavior simulation, we need empirical, or ‘real’, data about both the environment (mall) and the shoppers. In the literature review, we found only one paper that gathered empirical data about the shopper in a mall (Ruiz et al., 2001). The authors focuzed on the non-spatial data regarding shoppers and did not consider spatial aspects, neither about the shoppers, nor about the environment (the mall). In order to geosimulate the shopping behavior in a mall, we need spatial data about both the shoppers and the mall. For this reason, we decided to gather our own data in order to feed the simulation models. With regards to the simulation environment model (the mall), we use data stored in a geographic information system (GIS), and other data sources related to the mall. It has been proven that GIS is very efficient at storing non-spatial and spatial data about geographic environments (Benenson and Torrens, 2001). Regarding the model of the simulation’s main actor (the shopper), we use data gathered from real shoppers in the mall using the questionnaire technique. This technique seems to be very efficient at gathering detailed information about the shopper, but it is also very expensive in terms of time and effort. We gathered data at two malls: Square One in the Toronto area (October 2003) and Place De La Cité in Quebec City (July 2003). For this dissertation, we use the Square One mall as our simulation’s test case.

This section aims to present how we gathered the data which feeds the shopping behavior simulation models discussed in the previous chapter. In the previous chapter we presented two kinds of agents that compose the simulation model: passive agents that compose the simulation environment (mall) and active agents that represent the main actors of the simulation (shoppers). In order to feed the simulation models, we need to gather two types of data: the data related to the simulation environment (the mall), and the data related to the shoppers. These two kinds of data are presented, respectively, in the following two sub-sections.

In the literature about shopping behavior, we noticed that there are few research works that deal with empirical data about shoppers and shopping behavior in a mall. Furthermore, the majority of existing documents focus on the non-geographic data with regards to shoppers (gender, age, etc.). There is no research involving spatial data neither about shoppers, nor about shopping behaviors in malls (movements, itineraries, etc.). However, since we deal with geosimulation, such data is extremely important to us, and due to this lack of works dealing with empirical geographic data, we decided to gather our own data about shoppers and shopping behavior from spatial Square One mall. This information is both spatial and non-spatial. In order to gather data about shoppers through a questionnaire, we went through a number of sub-steps, which are presented in the following points.

- Preparing the questionnaire structure : In this step we prepared the structure of the questionnaire that was used to gather information about shoppers. The questionnaire aimed to feed the shopper model that was presented in the previous chapter. Hence, in the questionnaire preparation, we are based upon the simulation models’ structure presented in the previous chapter. What’s more, as mentioned in the previous chapter, the shopper model contains a variety of variables that are necessary for the simulation. However, we cannot gather data about all the variables belonging to this model because (1) if we wanted to gather information about all variables, the questionnaire would be much too long and boring. It is difficult to convince a shopper to interrupt his/her shopping trip in order to fill out a long questionnaire, therefore, the questionnaire must be tight and efficient , and (2) some of the variables belonging to the shopper model are psychological or social in nature, about which we are unable to gather empirical data, hence, we needed to exclude such variables from the questionnaire .

Details of the questionnaire, used to gather data about Square One shoppers are presented in Annex A of this thesis. In the following paragraphs, we briefly present the main sections which compose the questionnaire structure:

The questionnaire is divided into two main parts. The first part aims to gather data about individual shoppers (the individual questionnaire), while the second part focuzes on the groups of shoppers (group questionnaire).

The individual questionnaire is composed of six main sections, which are:

■ Section 0: This section aims to identify the questionnaire, describe the project to the respondent, explain the confidentiality instructions, present the recompense and award instructions, and present the general instructions of the questionnaire. This section is read by the questionnaire interviewer.

■ Section 1: It contains questions regarding the respondent’s demographic profile: gender, age group, occupations, sectors of employment, marital status, life modality, cohabitation, and postal code. The questions asked in this section are useful to better know the respondent. To complete this section we asked the respondent about his/her frequency of visits, from where he/she came (origin), the mode of transportation used to come to the mall, and the household income. This section is filled out by the respondent. The answers of the questions asked in this section will be used to feed the demographic variables (age, gender, occupations, etc.) shopper’s model presented in Chapterc 4.

■ Section 2: It contains an oral interview concerning the patron’s general and specific mall-visit objectives. This section requires pre- and post-shopping interviews during which the surveyor establishes the initial goals of the subject, and whether or not those goals were accomplished.

■ Section 3: Also in an interview mode, the surveyor examines the individual’s spatial usage of the two levels of Square One mall. In this section, we record the patron’s parking location, point of entrance, usual shopping itineraries, and shopping area preferences.

■ Section 4: The fourth segment of the individual questionnaire investigates the habits, preferences and interests of the respondent. These questions can be related to mall usage and mall environment (atmosphere).

■ Section 5: Also in an interview mode, the surveyor asks the respondent about his/her reactions and emotional states when he/she is found in some specific shopping situations.

The group questionnaire is composed of three sections, which are:

■ Section 0: This section aims to identify the group questionnaire, and present its general instructions. This section is read by the questionnaire interviewer.

■ Section 1: It contains some questions about the group identification, the persons who compose it, etc.

■ Section 3: The last section contains an oral interview concerning the group’s usual shopping itineraries in Square One mall.

- Conducting the survey : In order to conduct the survey, the surveyors selected three entrances based upon customer flow meters and suggestions from mall management as to which entrances showed the highest activity. The survey was carried out in two parts, one upon the individual’s arrival into the mall, and the other upon his/her exit. This enabled the surveyors to collect information regarding the individuals’/groups’ planned activities, as well as unexpected activities which were done during the mall visit.

The goal of the overall survey process was to gather data about a typical week at the Square One shopping mall center. As a result, the questionnaire was administrated over the course of four days, two of which were during the week and the other two were on the week-end. The survey was carried out during the full duration of the mall’s operating hours in order to capture the whole variety of mall users throughout the course of a normal day.

After four days, we gathered 390 completed individual questionnaires. However, the surveyors did not collect any data about groups of shoppers. During the survey carried out at Place De La Cité mall in Quebec City, we got few completed group questionnaires. This was not enough to carry out meaningful statistical analyses.

- Digitalizing survey data : Until this sub-step, the gathered data existed in paper format. To be used by a computer simulation prototype, this data needed to be digitalized. The digitalization process involves the following steps:

The design of the database structure, which contained the gathered information about the shoppers : The shopper’s database structure is based upon the structure and content of the questionnaire. Since we have two kinds of questionnaires (for individuals and groups), we have two parts in the shopper’s database structure: one for the individual shoppers and the other for the groups. The structure of this database was designed using Microsoft Access software. A brief description of this structure (the tables’ names) is presented in Fig 6.5 and Fig 6.6 for the individual questionnaire, and Fig 6.7 for the group questionnaire.

The data input into the database : In order to input the gathered geographic and non-geographic data into the database, we used a specific user-friendly software that we developed using Microsoft Visual Basic 6.0. Fig 6.8 presents one of the screenshots[3], which is used to input non-geographic data about a shopper (demographic data), and Fig 6.9 presents an example of screen used to input geographic data (itineraries of the shopper).

The gathered data aims to feed the shopping behavior multiagent model presented in Chapter 5. To gather data about the shoppers we used a survey. The structure and the content of this survey were defined based upon the initial version of the shopping behavior presented in Chapter 4. In this survey, we were not able to collect data about all the variables presented in the initial version of the model for the following reasons:

- Some variables are psychological, such as personality and values, and we did not find an easy and efficient way to gather data related to these variables;

- The survey is conducted during the respondent’s shopping trip, and must take little time[4].

The spatial and non-spatial data gathered during the survey must now be analyzed for different reasons:

- they can be used to understand how the shoppers use the environment (mall) and interact with its spatial elements when they perform their shopping behavior;

- they can be used to define the patterns of shopping behavior in a mall. These patterns will be used by the simulation engine of the simulation platform which will be used to execute the shopping behavior simulation models;

- they can give us an idea about the frequency of usage of the mall’s elements (doors, stores, corridors, etc.) by the shoppers;

This section aims at presenting specific analysis results, which are performed on the input data using an analysis tool that we developed for this purpose. Since we deal with geosimulation, spatial data has the same importance as non-spatial data. Referring to the two types of gathered data (spatial and non-spatial), we present two kinds of analysis: spatial and non-spatial.

The non-spatial analysis is related to the non-spatial gathered data. This kind of analysis is important, because it shows the distribution of the gathered data based upon the non-spatial variables of to the multiagent geosimulation model.

This sub-section briefly outlines some non-spatial analysis results of particular variables belonging to the model of the shopper agent. These results can inform us about the nature of shoppers that frequent the Square One mall. This will help us to define the categories of shoppers frequenting the mall (e.g., younger, older, students, etc.) and their typical behavior (e.g., browsing, making exercice, visitic specific stores, etc.). In the following points we present specific analysis results based on some non-spatial variables coming from the shopper model. The analysis results for all the variables belonging to the shopper’s model are not presented in this manuscript.

- The variable ‘Gender of the shoppers’ : As Table 6.1 and Fig 6.10 illustrate, there is a relatively even split of respondents between genders with 55% female, and 44% male.

    • The variable ‘Age group of the shoppers’ : As shown in Table 6.2 and Fig 6.11, the largest age group by percentage was that of 18-25 years, representing 28% of the questionnaire respondents. People younger than 17 years of age made up about 17% of the survey, and those aged between 26 and 35 years represented 24%. Respondents between 36 and 50 years old represented 21% of the interviewees. What’s more, it is interesting to notice that there were only 4% over the age of 66 years, and 6% who are between 51 and 65 years.

We also analyzed the following variables belonging to the demographic profile of the shopper: household income, occupation, sector of employment, mean of transportation, origin (from where the shoppers come), the life modality, and marital status, etc. The analysis results of these variables are not presented in this dissertation.

The variable ‘General objectives (planned objectives)’ : Table 6.3 and Fig 6.12 show that the majority of respondents come to the mall in order to do some window shopping (51%), or to accompany a person (62%).

The variable of ‘Shopping period in the day’ : Table 6.4 and Fig 6.13 demonstrate that the most preferred periods in the day for shopping are in the afternoon and evening.

We also analyzed the following variables related to the shopper’s preferences: shopping day, type of corridors in the mall, parking type, used parking, lighting type, display and decoration colors, mall visit frequency, and musical choices.

- The interest variable : Table 6.5 shows the analysis results of the interest variable of the respondents. As demonstrated in this table, the majority of respondents are always interested in travels/tourism, cultural activities, music, cinema/television, etc. On the other hand, they are not always interested in society games or land vehicles.

Non-spatial analysis results presented in this sub-section are related to some variables related to the individual shopper model. These analyses involve one variable at a time. These results help us to create some categories of shoppers visiting Square One mall. For example, as mentioned above, the analyses results related to the non-spatial variable ‘General objectives (planned objectives)’ related to the shopper model inform us about the categories of shoppers visiting Square One mall. Most of these shoppers are browsers which come to the mall to do some window shopping, or individuals that accompagny another person. These results are also used as a base for multi-variables analysis which is discussed in sub-sections 6.4.3 and 6.4.4 which is important to define the shopping behavior patterns. The definition of these patterns is discussed in the next chapter.

Since we deal with geosimulations, spatial and geographic data should be carefully analyzed. The majority of spatial data belongs to the environment model. This analysis aims to examine the frequencies of usage of the environment’s spatial entities (parking, doors, corridors, shopping areas, etc.).

In this sub-section, we present some analysis results related to some spatial variables (or entities) belonging to the environment model such as the parkings and the entrance doors. These analysis results inform us about the frequency of usage of these spatial variables by the shopper. These analyses, which involve one spatial variable at a time, provide the basis for multi-variables analyses that are important to define the shopping behavior patterns. The definition of these patterns is discussed in the next chapter.

- The ‘parking’ variable : This analysis shows the percentage use of parking areas by shoppers. Table 6.6 and Fig 6.14 indicate that the parking areas on first floor of Square One are not excessively frequented by shoppers. On the other hand, the parking areas on second floor are used very frequently (see Table 6.7 and Fig 6.15). These results are presented cartographically in Fig 6.16 and Fig 6.17. In these figures, the numbers in the circles identify the parking areas and the colors are proportional to the number of shoppers who use the parking areas. The higher the number, the darker the color

- The entrance door variable : This analysis shows the percentage of use of the mall’s entrance doors by shoppers. The analysis results in Table 6.8 and Fig 6.18 indicate that the most frequented entrance doors on first floor are doors 10 (32%) and 0 (24%). These results are presented cartographically in Fig 6.19. Here, the numbers in the circles identify the doors, and the colors are proportional to the flow of shoppers who reach the mall by this entrance. If the flow increases, the color of the circle becomes darker.

Table 6.8. The entrance door variable (first floor of Square One mall).

In the analysis results presented above, we deal with one variable at a time. In order to see the effect of each variable on the others, we analyzed the combination of several variables. This can help us to define more detailed behavior patterns for the system to be simulated. For example, for the shopping behavior in a mall, two-variables analysis results, related to the variables general objectives and gender , can inform us that the majority of browsers are females. Based upon this information, we can define two detailed shopper categories (browsers females and browser males) and we define a specific pattern for each category of shoppers. Multi-variables analysis is more interesting if we analyze more than two variables. For example, a three-variables analysis ( general objectives, gender , and preferred music ) can inform us that the majority of the females browsing the mall like classical music. Hence, we can design a more detailed classification of the shoppers visiting the mall based on these variables.

In this sub-section, we present an example of analysis, in which we combine non-spatial variables related to the shoppers: gender and age group. Table 6.9 and Fig 6.20 show that among the respondents who are between 13 and 17 years of age (teenagers), the majority are female (47/65 or 72%), while males represent only 27% (18/65). On the other hand, we can see that the percentage of females who are between 18 and 35 years of age is almost equal to the percentage of males in this age category.

In geosimulation, analyzing the combination of the non-spatial variables related to the model of the system to be simulated and the spatial variables related to the simulation environment model is very important because we can determine how spatial variables can influence non-spatial one. For example, in our shopping behavior case, if we combine the non-spatial variable gender of the shopper with the spatial variable entrance door of the simulation environment (mall), we can identify precisely which kinds of shoppers attend the entrance door. Then, and based upon this information, we can define more detailed spatial categorization of the shoppers by entrance door.

The spatial analysis results presented in Table 6.8, Fig 6.18, and Fig 6.19 show us that door number 0, on first floor, is frequented by 97 shoppers, while door number 10 is frequented by 126 shoppers. Nevertheless, this kind of analysis cannot answer the following question: Who are these shoppers? In order to answer this type of question, we need to combine the spatial variables of the environment (entrance doors, parking areas, etc.) with the non-spatial variables related to the shoppers (age group, gender, etc.).

In order to know what is the gender of those shoppers frequenting the door number 0, we made a non-spatial/spatial analysis of the combination of two variables: the variable entrance door number 0 (spatial analysis) and the variable gender of the shoppers (non-spatial analysis). The analysis results in Table 6.10 and Fig 6.21 demonstrated that among the 97 shoppers who frequented door 0, 60 were female and 37 were males.

We can analyze a combination of more than two variables in order to know, in more detail, who are the shoppers that frequent a specific area in the environment. For example, we can combine the variables: gender, age group (non-spatial variables), and entrance door (spatial variable) to know exactly who are the shoppers that frequent door number 0. The results presented in Table 6.11 and Fig 6.22 show that the majority of females who frequent entrance door 0 (60 females) are between 13 and 17 years of age (20/60).

Among these 60 female clients, we can do multidimensional analysis to find the number of females from each age group, giving us the following results:

The shoppers’ database contains numerous spatial and non-spatial variables that can be combined together. These variables are summarized in Table 6.12.

The analysis results of the simulation input data gathered during this step will be used in the sixth step of our method in order to define some shopping behavior patterns and to initialize the execution of the geosimulation (see Chapter 7 of this dissertation).

This chapter presented the step in which we gather and analyze the input data used to feed the simulation models designed in the previous step. We claim that the geographic data is a key element in geosimulation. Hence, spatial or geographic data has the same importance as non-spatial variables. For this reason, we need to gather and analyze spatial and non-spatial data related to the phenomenon to be simulated and to its simulation environment. Since geosimulation is a young field, we did not find in the literature any method’s or and approach’s step that deal with spatial data analysis of the simulation input. Referring to the shopping behavior simulation models presented in the previous chapter, we collected empirical data about the simulated environment (the mall) and information about the simulation’s main actor, which is the shopper. In order to be used by a computer simulation, the gathered data was digitalized and stored in a database. Moreover, in order to find some shopping behavior categories and patterns which can be useful for the simulation, we analyzed the gathered data. The important part of this study is the analysis of the combination of non-spatial data related to the shopper’s model and spatial data related to the mall’s model. This analysis can give us an idea about the frequency of use of the mall’s geographic elements.

In this chapter, we presented an interesting step in which we collect and analyze data before using it to feed the geosimulation models. This step is, generally, neglected by research works dealing with simulation methods and approaches such as (Anu, 1997), (Fishwick, 1995), (Allen et al., 2001), (Groumpos and Merkuryev, 2002), (Drogoul et al., 2002), (Ramanath and Gilbert, 2003), etc. Introducing such a step in our method can be considered as an original work in the field of simulation. What’s more, (1) developing a software to digitalize non-spatial and spatial simulation data at once and (2) analyzing these two kind of data can also be considered as original work. It is interesting to note that the preliminary findings derived from the collected data and presented to the managers of Place De La Cité mall raised a lot of interest.

In this chapter we presented a survey method to collect spatial and non-spatial real data about a phenomenon for geosimulation. This method which is illustrated using the shopping behavior case study can be considered as a contribution, because we did not find any similar method or technique that can be used to collect spatial empirical data about a specific phenomenon which operates in geographic context.

Another contribution consists in the development of a technique that can be used to analyze uni-variable or multi-variables spatial and non-spatial input data of a geosimulation.

The next chapter aims at presenting the implementation of the shopping behavior simulation models presented in chapter 5, using the gathered data presented in this chapter.



[3] The software contains about fifteen windows which are used to record the collected data into the shopper’s database.

[4] Actually, the survey tooks about 30 minutes to interview a shopper. This is a long survey duration, and we are surprised by the cooperation of the shoppers who answered the questions.

© Walid Ali, 2006