• No se han encontrado resultados

ANÁLISIS FINANCIERO PARA EL ESCENARIO PESIMISTA

11. ANÁLISIS FINANCIERO

11.4. ANÁLISIS FINANCIERO PARA EL ESCENARIO PESIMISTA

characteristics

Let us make a very standard exercise in urban geography: depicting the structure of a territory by clustering places that look alike in terms of a set of

selected characteristics Xi, reminding the so called “factorial ecology”, a

well-known technique anchored in early quantitative geography and urban

modelling (Janson, 1980; Lebowitz, 1977). The objective is here to describe

mobile phone calls?

Figure 7.11. Zoom on Brussels (A) and Ottignies Louvain-la-Neuve (B) where locations on

the map are: 1 the Ixelles cemetery, 2 is the Eugène Flagey Place, 3 the Grand Place, 4 VW- Audi car factory (Forest), 5 Brussels Airport (Zaventem), 6 and 7 Sport centres, 8 Concert Hall (Forest National), 9 Ottignies, 10 family residential area, 11 student residential area, 12 Place des Sciences (party place) and 13 Industrial zoning areas (Parc Einstein, Parc Fleming and Axis Parc).

more specifically to see if Brussels and its suburbs emerge in the constellation

of smaller regional centres and periurban realities (Riguelle et al., 2007;

Servaiset al.,2004).

The characteristics of the places are provided by the latest census (Census-

2011, 2016) at the level of the statistical sections (Statbel, 2019). Let us

remind that each municipality is statistically divided into statistical sections and further sectors. Sections hence correspond to the aggregation of statistical sectors that are the smallest level of aggregation for which data are officially available in Belgium. Statistical sections were preferred to sectors because of their larger size, avoiding too many small/missing numbers. The studied area counts 982 sections among which 11 have missing values for at least one variable or count no inhabitant; we end up with 971 statistical sections.

A selection of six variables is used to characterize the socio-economic

realities and the urban environment (see Table 7.2). Their choice is largely

constrained by their availability at the section level and well inspired by

theory. These variables are: (1) population density (dpop), (2) the percentage

of youngsters (< 15y) in the total population (lessthan15 y.), (3) the

unemployment rate (Unemploym), (4) the percentage of commuters working

in a section but residing in another municipality than that of the section

(commut.in), (5) the proportion of dwellings built before 1946 (builtbef.1946)

(rentedresid). Unfortunately, it was impossible to use income data due to the too many missing values observed (small number of inhabitants) and it was

decided to apply our analyses on specific proxy variables, like the

unemployment rate and the percentage of dwellings occupied by a tenant (Timms, 1970). dpop, commut.in and builtbef.1946 describe the urban environment where people live as well as the economic activities, while the other variables are used as indicators of the social conditions of the

households (Janson, 1980; Sapena et al., 2016; Timms, 1970). Variables are

standardized and Table 7.3 shows that the variables are positivity correlated

at the exception of commut.in that are negative (or more weakly) with the

proportion of youngsters and that of old housings.

A Ward classification is applied in order to group the 971 sections into a small number of clusters based on their resemblance on the six selected

variables (Rokach and Maimon,2005;Ward,1963). Following the method used

in Section7.3(small values for Intra-Clusters Variances and the Entropy while

the Dunn and Calinski-Harabasz is maximal), the optimal number of clusters

is here 4. Figure 7.12, as well as Figure 7.15, summarize the composition

and the delineation of the four clusters. Clusters 1 and 2 reflect quite well the urban structure within the former province of Brabant, where Brussels sprawls out of the limits of its administrative border (BCR). Cluster 1 (red on

Figure7.12) is the smallest cluster (98 sections) that mainly corresponds to the

historical urban centre of Brussels (De Keersmaecker et al., 2003; Eggerickx

et al., 2010). It is densely built, and characterized by a high proportion of

old housings, mostly rented, and often quite deprived (Van Criekingen,2006).

Only three out of these 98 statistical sections are not located in the BCR.

Cluster 2 (237 sections, orange in Figure7.12) has a different population profile:

less densely built, higher percentage of jobs occupied by workers coming from further away. People are commuting from other municipalities to join their place of work located within and around the major cities as well as within

economic zoning, mainly located in the suburbs (Ermanset al.,2018;Verhetsel

et al., 2018). Regional centres emerge (Leuven, Nivelles, Wavre, Ottignies) within less urbanized places in the North than in the South, but these orange sections also correspond to former municipal centres.

Clusters 3 and 4 are the two largest clusters and correspond to places located further away from Brussels and hence less urbanized: Cluster 3 (263 sections) is largely represented in the southern part of the province, while Cluster 4 (373 sections) dominates the Northern urban landscape. Both clusters correspond to places with a lower population density and a more periurban landscape (Thomas et al., 2007). The main difference between Clusters 3 and 4 is that the dwellings are more recent in Cluster 4 as could be expected from former housing analyses and by different land use policy between Flandres and

mobile phone calls?

We obtain a very clear centre periphery structure (Cluster 1 and partially Cluster 2) around Brussels, and a periphery divided into a North/South organization: the Flemish periurbanisation (Cluster 4) and the Walloon one (Cluster 3). However some green coloured sections are found in the North and orange in the South, The Flanders/Wallonia difference is well marked but not clear-cut. Finally, the regional city of Leuven only slightly emerges.

Figure 7.12. Clusters of statistical sections based on socio-economic variables.

If communities detected with phone calls (see Section 7.3.2) revived the

Hoyt sector model, socio-economic basins and more anchored in the Burgess model (centre periphery, modified by political realities) and hence reminding the complexity of a urban territory where several spatial structures are layered.