Data on Healthy Food Accessibility in Amsterdam, The ... - MDPI

Download PDF
5 downloads 5414 Views 3MB Size Report
Comment
Jan 26, 2017 - and the possibility to validate results, to apply alternative analytical ... The Dutch cadastral data Basisregistraties Adressen en Gebouwen [14] and ArcGIS Online ... In order to perform statistical analyses, the initial classes were ...
data Data Descriptor

Data on Healthy Food Accessibility in Amsterdam, The Netherlands Marco Helbich 1, * and Julian Hagenauer 2 1 2

*

Department of Human Geography and Spatial Planning, Utrecht University, 3584 CS Utrecht, The Netherlands Leibniz Institute of Ecological Urban and Regional Development, 01217 Dresden, Germany; [email protected] Correspondence: [email protected]; Tel.: +31-30-253-2017

Academic Editor: Xinyue Ye Received: 16 November 2016; Accepted: 22 January 2017; Published: 26 January 2017

Abstract: This data descriptor introduces data on healthy food supplied by supermarkets in the city of Amsterdam, The Netherlands. In addition to two neighborhood variables (i.e., share of autochthons and average housing values), the data comprises three street network-based accessibility measures derived from analyses using a geographic information system. Data are provided on a spatial micro-scale utilizing grid cells with a spatial resolution of 100 m. We explain how the data were collected and pre-processed, and how alternative analyses can be set up. To illustrate the use of the data, an example is provided using the R programming language. Data Set: http://www.mdpi.com/2306-5729/2/1/7/s1 Data Set License: CC-BY Keywords: data; reproducible research; food deserts; health inequalities; accessibility measures; Amsterdam (The Netherlands)

1. Introduction Spatial accessibility to healthy food is important for people’s health [1]. In that respect, supermarkets play an essential role by offering healthy and fresh foods at more competitive prices than smaller grocery stores or convenience stores [2]. However, supermarket access is not constant, but varies significantly across cities, resulting in dietary inequalities across urban neighborhoods. The body of knowledge thus far suggests that particularly people residing in socially-distressed neighborhoods (i.e., having a low socioeconomic status) as well as those neighborhoods where predominantly ethnic minorities live have poorer spatial supermarket accessibility. Such areas are often denoted as “food deserts” [3]. While food deserts seem to be omnipresent in the U.S., evidence concerning their existence in Canadian or European cities is mixed and far from conclusive [1]. Reasons for divergent findings include the applied methodology, which is mainly based on geographic information systems (GIS) to compute accessibility indicators, and the applied statistical models [4]. Present studies are often conceptually simple, applying a single accessibility measure on a less detailed analytical scale (e.g., administrative units). Therefore, multidimensional accessibility indicators combining proximity to, and density and variety of, supermarkets are suggested [5–7]. Although promoting a straightforward operationalization, food deserts are frequently identified by means of descriptive approaches (e.g., quartiles), disregarding that both accessibility and neighborhood characteristics are key for food desert mapping which calls for multivariate data clustering [4]. Finally, to the best of our knowledge, the conducted studies do not make the underlying research data (e.g., primary data, secondary data, and derived measures) available to the public, even

Data 2017, 2, 7; doi:10.3390/data2010007

www.mdpi.com/journal/data

Data 2017, 2, 7

2 of 10

though the reproducibility of findings on which knowledge is built is imperative, and a fundamental aspect in scientific investigations. Brunsdon [8] critically highlights several benefits when methods and data repositories are shared. The benefits include clear documentation, transparency concerning pre-processing, and the possibility to validate results, to apply alternative analytical approaches, to serve as a basis for follow-up studies, etc. All these issues will ultimately lead to more reliable research. This data descriptor addresses the aforementioned research gaps by describing in detail and sharing the data related to the research article Food Deserts? Healthy Food Access in Amsterdam [9]. It describes both the data collection and the procedures used in pre-processing the data, and gives an overview of how the data can be used. Note that the interpretation of the results is given in the companion article. The provided data is not only relevant to map healthy food accessibility but is also of relevance for other studies dealing with, for example, the analyses of health behavior and can be linked to on-going area-based or register studies in Amsterdam. 2. Data Description Table 1 summarizes key characteristics of the dataset. Table 1. Metadata specification. Key Features

Description

Subject area

Health, nutrition, geography, transportation

Data source location

Amsterdam, The Netherlands

Data acquisition

Derived attributes and official data (Statistic Netherlands)

Type and format

R object (SpatialPolygonsDataFrame), ESRI™ shapefile

Spatial resolution

cells with 100 m widths

Dimension

5242 × 8

Projection and reference system

EPSG code: 28992

Attributes Proximity (PROX)

Numeric, distance to the closest supermarket from each cell (in meters)

Density (DENS)

Numeric, number of stores within a 1000 m street network buffer around each cell

Variety (VARI)

Numeric, mean distance to three supermarkets of three different chains from each cell (in meters)

Ethnicity (NATI)

Numeric, proportion of native Dutch within a cell in the year 2014 (converted to the following numeric values: 5 = 90%, 4 = 75%–90%, 3 = 60%–75%, 2 = 40%–60%, 1 = 40%)

Housing (HOUS)

Numeric, average housing price per cell in the year 2011/12 (in €1000)

ID

Unique identifier

Version 1 1

1.0

In case the data will be updated, the version number will be changed. Older versions will be archived.

3. Materials and Methods 3.1. Study Area and Analysis Scale The study area was the city of Amsterdam, The Netherlands. The city is located at 52◦220 N, 4◦530 O. Figure 1 shows the location of the study area. We selected Amsterdam because the health monitor [10] reported distinct differences in overweight and obesity prevalence. For example, 40% of the residents are overweight, and 75% of the adults do not consume the recommended amount of fruit and vegetables. Significant spatial variation in overweight prevalence exists as well. With 22% pronounced overweight, prevalence can be found in the central areas and the rates increase even further in the northern parts of Amsterdam. In contrast to most food desert studies, we tried to circumvent methodological complications arising from the application of census areas (e.g., an uneven size). In order to go beyond administrative units, we overlaid the study area with a grid in which each cell had a spatial resolution of 100 × 100 m. Thus, information is available for 5242 cells in total. Note that the data provided here only includes cells where people reside; cells without a residential population were queried and excluded from further analyses. Furthermore, an ID (e.g., E1281N4931) introduced by Statistic Netherlands [11] was

Data 2017, 2, 7 Data 2017, 2, 7

3 of 10

3 of 10

excluded from further analyses. Furthermore, an ID (e.g., E1281N4931) introduced by Statistic

attached to each cell, a straightforward with administrative data. Theother grid cells Netherlands [8] allowing was attached to each cell,linkage allowing a other straightforward linkage with are provided as an ESRI™ shapefile and R data object (see Section 4). administrative data. The grid cells are provided as an ESRI™ shapefile and R data object (see Section 4).

Figure 1. Study area.

Figure 1. Study area.

3.2. Data Sources and Pre-Processing

3.2. Data Sources and Pre-Processing 3.2.1. Supermarket Data

3.2.1. Supermarket Data

The initial search for all supermarket chains operating in The Netherlands was guided by an

overview published in Wikipedia [9] and the newspaper Levensmiddelen Krant [10]. Thewas number of by The initial search for all supermarket chains operating in The Netherlands guided stores perpublished supermarket chain located within unit of Amsterdam, those an overview in Wikipedia [12] and the theadministrative newspaper Levensmiddelen Krantincluding [13]. The number within a buffer zone of two kilometers thethe city, were collected. unit The consideration of a buffer of stores per supermarket chain locatedaround within administrative of Amsterdam, including zone was necessary to avoid edge effects for the accessibility measures. Due to theoretical those within a buffer zone of two kilometers around the city, were collected. The consideration of considerations, organic supermarkets and “to go” stores were disregarded. Each company’s webpage a buffer zone was necessary to avoid edge effects for the accessibility measures. Due to theoretical was queried to obtain the store addresses (i.e., the street name and the building number). A total of considerations, organic supermarkets and “to go” stores were disregarded. Each company’s webpage 144 supermarkets were identified during the data collection phase in November 2015; of them, 122 was queried to obtain theadministrative store addresses the street name the building number). about A total of are located within the area(i.e., of Amsterdam. Table and 2 provides some information 144 supermarkets during the data collection phase in November 2015; of them, 122 are the stores. Thewere Dutchidentified cadastral data Basisregistraties Adressen en Gebouwen [11] and ArcGIS Online located within the administrative area of Amsterdam. Table provides some information were then used to convert the individual store addresses into2 geographic coordinates, which about were the then projected the local coordinate system (i.e.,Adressen EPSG codeen28992). A detailed of the stores. The Dutch onto cadastral data Basisregistraties Gebouwen [14]description and ArcGIS Online projection is given in Section 4. were then used to convert the individual store addresses into geographic coordinates, which were then projected onto the local coordinate system (i.e., EPSG code 28992). A detailed description of the Table 2. Supermarket chains. projection is given in Section 4. Chain Number of Stores Chain Number of Stores Table chains. Albert Heijn 79 2. Supermarket Coop 3 Dirk 15 Plus 4 Chain Number of Chain Number of Stores Jumbo 15Stores Spar 3 Lidl Dekamarkt 2 Albert Heijn 7910 Coop 3 Dirk Jumbo Lidl Aldi Deen

15 15 10 5 6

Plus Spar Dekamarkt C1000 Boni

4 3 2 1 1

Data 2017, 2, 7

4 of 10

3.2.2. Accessibility Measures For the accessibility measures, the coordinates of the centroid of each cell, serving as origin, were computed, while the supermarket locations served as destinations. The accessibility indicators were calculated on the basis of the street network provided by ESRI (version 2008) as input and a function iterated over all origins (cells). Based on a literature review (e.g., [4,5]), three complementing supermarket accessibility measures were considered. The first indicator is based on the network distance (in meters) from cell i to the closest supermarket j of any chain (proximity measure). For the second measure, we first computed a street network buffer (service area) of 1000 m around each cell centroid, and then applied GIS-based point-in-polygon analyses to determine the number of available stores within this area (density measure). The threshold distance is based on a review of the literature and represents a 12-min walk for an adult [4]. The final accessibility measure differentiates between supermarket chains, and represents the mean network distance (in meters) from each centroid i to the three nearest supermarkets j from k different chains (variety measure). The variety measure considers that different chains offer different products. To derive these measures, the ArcGIS 10.3 network analyst extension was used with the centroids as incidents and the supermarket locations as facilities. For all analyses, all the routing restrictions were disabled (e.g., one-ways). 3.2.3. Neighborhood Data We extracted neighborhood information for two variables for each cell from the raster dataset (vierkanten) maintained by Statistics Netherlands (www.cbs.nl) [11]. The first variable represents ethnicity and is based on the municipal personal records database (Structuurtelling Gemeentelijke Basisadministratie). The variable ethnicity represents the proportion of autochthons, that is, the proportion of people whose parents were born in The Netherlands, irrespective of their country of origin [11]. The variable was originally coded as follows: (1) ≥90% autochthons; (2) 75%–90% autochthons; (3) 60%–75% autochthons; (4) 40%–60% autochthons; and (5) library(sp); library(rgdal) > library(sp); library(rgdal) library(spdep); library(kohonen) > library(spdep); library(kohonen) library(RColorBrewer) > library(spdep); library(kohonen) > library(RColorBrewer) > library(RColorBrewer) > # set workspace > # set workspace setwd("D:/workspace") > # set workspace > setwd("D:/workspace") > setwd("D:/workspace") getwd() > getwd() > getwd() [1] "D:/workspace" [1] "D:/workspace" > # "D:/workspace" Download zip from the web (link needs to be adjusted) [1] > # Download zip from the web (link needs to be adjusted) > # Download zip from the web (link needs to be adjusted) download.file("http://ADD_LINK/data_amsterdam.zip", "data_amsterdam.zip") > # download.file("http://ADD_LINK/data_amsterdam.zip", "data_amsterdam.zip") > # download.file("http://ADD_LINK/data_amsterdam.zip", "data_amsterdam.zip") unzip file > # unzip file unzip("data_amsterdam.zip") > # unzip file > unzip("data_amsterdam.zip") > unzip("data_amsterdam.zip")

The ESRI shapefile is then read from the workspace and the projection is queried. The bounding The ESRI shapefile is then read from the workspace and the projection is queried. The bounding The shapefile is then from the workspace and the projection is queried. The bounding box of theESRI extended study area read is also plotted. box of the extended study area is also plotted. box of the extended study area is also also plotted. plotted. > # load shapefile > # load shapefile data # load > data data # save(data,file="data_amsterdam") > # save(data,file="data_amsterdam") projection of the data > # save(data,file="data_amsterdam") > # projection of the data > # projection of the data proj4string(data) > proj4string(data) > proj4string(data) [1] "+proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 + [1] "+proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 + [1] "+proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 + k=0.9999079 +x_0=155000 +y_0=463000 +ellps=bessel +units=m +no_defs" k=0.9999079 +x_0=155000 +y_0=463000 +ellps=bessel +units=m +no_defs" k=0.9999079 +x_0=155000 +y_0=463000 +ellps=bessel +units=m +no_defs" > # bounding box > # bounding box bbox(data) > # bounding box > bbox(data) > bbox(data) min max min max min 130800.0 max x 111969.8 x 111969.8 130800.0 y 111969.8 477900.0 130800.0 493142.9 x y 477900.0 493142.9 y 477900.0 493142.9

Additional summary summary statistics of of the variables variables are reported reported below. Additional Additional summary statistics statistics of the the variables are are reportedbelow. below. Additional summary statistics of the variables are reported below. > # show attributes > # show attributes > # show attributes names(data) > names(data) [1] "DENS" "ID" "PROX" "NATI" "HOUS" "VARI" > names(data) [1] "DENS" "ID" "PROX" "NATI" "HOUS" "VARI" [1] "PROX" "NATI" "HOUS" "VARI" > # "DENS" summary"ID" attributes > # summary attributes > # summary attributes summary(data[, c("NATI", "HOUS", "PROX", "DENS", "VARI")]) > summary(data[, c("NATI", "HOUS", "PROX", "DENS", "VARI")]) > summary(data[, c("NATI", "HOUS", "PROX", "DENS", "VARI")]) Object of class SpatialPolygonsDataFrame Object of class SpatialPolygonsDataFrame Coordinates: Object of class SpatialPolygonsDataFrame Coordinates: Coordinates: min max min max x 111969.8 min 130800.0 max x 111969.8 130800.0 x y 111969.8 477900.0 130800.0 493142.9 y 477900.0 493142.9 Is477900.0 projected: TRUE y 493142.9 Is projected: TRUE proj4string : TRUE Is projected: proj4string : proj4string : +lat_0=52.15616055555555 +lon_0=5.38763888888889 +k=0.9999079 + [+proj=sterea [+proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 +k=0.9999079 + [+proj=sterea +lat_0=52.15616055555555 +lon_0=5.38763888888889 +k=0.9999079 + x_0=155000 +y_0=463000 +ellps=bessel x_0=155000 +y_0=463000 +ellps=bessel x_0=155000 +y_0=463000 +ellps=bessel

Data 2017, 2, 7 Data 2017, 2, 7 Data 2017, 2, 7 Data 2017, 2, 7

6 of 10 6 of 10 6 of 10 6 of 10

+units=m +no_defs] +units=m +no_defs] Data attributes: +units=m +no_defs] Data attributes: NATI HOUS PROX DENS VARI Data attributes: NATI HOUS PROX DENS VARI Min. :1.000 Min. : 35.0 Min. : 0.212 Min. : 0.000 Min. : Min. NATI :1.000 Min. HOUS : 35.0 Min. PROX : 0.212 Min. DENS : 0.000 Min. VARI : 104.3 Min. :1.000 Min. : 35.0 Min. : 0.212 Min. : 0.000 Min. : 104.3 1st Qu.:1.000 1st Qu.: 176.0 1st Qu.: 349.360 1st Qu.: 1.000 1st Qu.: 104.3 1st Qu.:1.000 1st Qu.: 176.0 1st Qu.: 349.360 1st Qu.: 1.000 1st Qu.: 707.3 1st Qu.:1.000 1st Qu.: 176.0 1st Qu.: 349.360 1st Qu.: 1.000 1st Qu.: 707.3 Median :2.000 Median : 228.0 Median : 560.659 Median : 2.000 Median 707.3 Median :2.000 Median : 228.0 Median : 560.659 Median : 2.000 Median :1067.7 Median :2.000 Median : 228.0 Median : 560.659 Median : 2.000 Median :1067.7 Mean :2.275 Mean : 273.9 Mean : 684.490 Mean : 2.654 Mean :1067.7 Mean :2.275 Mean : 273.9 Mean : 684.490 Mean : 2.654 Mean :1240.4 Mean :2.275 Mean : 273.9 Mean : 684.490 Mean : 2.654 Mean :1240.4 3rd Qu.:3.000 3rd Qu.: 305.0 3rd Qu.: 891.356 3rd Qu.: 4.000 3rd :1240.4 3rd Qu.:3.000 3rd Qu.: 305.0 3rd Qu.: 891.356 3rd Qu.: 4.000 3rd Qu.:1592.7 3rd Qu.:3.000 3rd Qu.: 305.0 3rd Qu.: 891.356 3rd Qu.: 4.000 3rd Qu.:1592.7 Max. :5.000 Max. :2675.0 Max. :7454.897 Max. :10.00 Max. :7525.8 Qu.:1592.7 Max. :5.000 Max. :2675.0 Max. :7454.897 Max. :10.00 Max. :7525.8 Max. :5.000 Max. :2675.0 Max. :7454.897 Max. :10.00 Max. :7525.8 > # show data structure > # show data structure > str(data@data) > show data structure > # str(data@data) 'data.frame': 5242 obs. of 6 variables: > str(data@data) 'data.frame': 5242 obs. of 6 variables: $ DENS: num 0 0 0 0 1 0 0 0 0 0 ... 'data.frame': $ DENS: num 0 5242 0 0 0obs. 1 0 of 0 0 6 0 variables: 0 ... $ ID : Factor w/ 5242 levels "E1119N4855","E1120N4854",..: 5103 5145 5158 ... $ 0 0 0 1 0 0 0 0"E1119N4855","E1120N4854",..: 0 ... $ DENS: ID : num Factor w/05242 levels 5103 5145 5158 ... $ PROX: 4819 5119 1314 1085 ... $ : num Factor w/ 5019 5242 "E1119N4855","E1120N4854",..: 5103 5145 5158 ... $ ID PROX: num 4819 5019 levels 5119 1314 1085 ... $ NATI: num 5 5 5 5019 2 5 3 3 4 1314 3 1 ... ... $ $ PROX: NATI: num num 4819 5 5 5 2 5 5119 3 3 4 3 1 1085 ... $ HOUS: num 523 438 249 210 241 265 247 236 ... $ 5 5 2 5 3 3 4 343 3 1 293 ... $ NATI: HOUS: num num 5 523 438 249 210 343 293 241 265 247 236 ... $ VARI: num 5217 5417 5517 2412 2183 ... 265 247 236 ... $ 249 2102412 343 293 $ HOUS: VARI: num num 523 5217438 5417 5517 2183241 ... $ VARI: num 5217 5417 5517 2412 2183 ...

Next, the input variables for the SOM are selected. These variables are scaled before the SOM Next, the the input input variables variables for for the the SOM SOM are are selected. selected. These These variables variables are are scaled scaled before before the the SOM SOM Next, topology (10 × 8) is set up. Thisfor grid serves as input for the SOMvariables algorithm. Next, the input variables the SOM are selected. These are scaled before the SOM topology (10 (10 × × 8) topology 8) is is set set up. up. This This grid grid serves serves as as input input for for the the SOM SOM algorithm. algorithm. topology (10 × 8) is set up. This grid serves as input for the SOM algorithm. > # select variables for som-based clustering > # select variables for som-based clustering > data.som as.matrix(data@data[, c("NATI", "HOUS", "PROX", "DENS", "VARI")]) > select # data.som # scale selected variables > data.som # scale selected variables > data.som.sc # scale selected variables > data.som.sc # create som scale(data.som) > data.som.sc # create som topology > som.grid somgrid(xdim = 10, ydim = 8, topo = c("hexagonal")) > create # som.grid set.seed(20082014) > som.grid set.seed(20082014) > # train som > set.seed(20082014) > # train som > som.res som(data=data.som.sc, grid=som.grid, rlen=100, alpha = c(0.05, > train # som.res som.res summary(som.res) + keep.data=TRUE, n.hood="circular") > summary(som.res) som map of size 10x8 with a hexagonal topology. > summary(som.res) som map of size 10x8 with a hexagonal topology. Training data included; dimension is 5242 by 5 som map of size 10x8 with a hexagonal topology. Training data included; dimension is 5242 by 5 Mean distance to the closest unit in the map: 0.3151507 Training data included; dimension is 5242 by Mean distance to the closest unit in the map:50.3151507 Mean distance to the closest unit in the map: 0.3151507

The key key strengthsof of SOMsare are theirrich rich visualizationcapabilities. capabilities. The SOM training progress The The SOM training progress in The key strengths strengths ofSOMs SOMs aretheir their richvisualization visualization capabilities. The SOM training progress in Figure 2 shows considerable improvements after a few iterations and convergence after The key of SOMs areimprovements their rich Theand SOM training progress Figure 2 shows considerable improvements aftervisualization a few iterations convergence after approximately in Figure 2 strengths shows considerable after a capabilities. fewand iterations convergence after approximately 50 training cycles. Moreover, the U-matrix [17] caniterations beinvestigate plottedand to investigate clusters, in Figure 2 shows considerable improvements after a few convergence after 50 training cycles. Moreover, the U-matrix [19] can be plotted to clusters, while the approximately 50 training cycles. Moreover, the U-matrix [17] can be plotted to investigate clusters, while the component planes are useful to discover correlations among the variables. Figure 3 depicts approximately 50 training cycles. Moreover, the U-matrix [17] can be plotted to investigate clusters, component planes areplanes usefulare touseful discover correlations among among the variables. FigureFigure 3 depicts two while the component to discover correlations the variables. 3 depicts two component planes asexample. an example. while the component planes useful to discover correlations among the variables. Figure 3 depicts component planes as an two component planes as an are example. two component planes as an example. > # set colors > # set colors > coolBlueHotRed set colors > # coolBlueHotRed coolBlueHotRed # plot som training progress alpha=alpha)[n:1] } > # plot som training progress > # plot som training progress

Data 2017, 2, 7 Data 2017, 2, 7

7 of 10 7 of 10

Data 2017, 2017, 2, 2, 77 Data

of 10 10 77 of

> plot(som.res, type="changes") > plot(som.res, plot(som.res, type="changes") type="changes") > Data 2017, 2, 7

7 of 10

> plot(som.res, type="changes")

> > > >

Figure 2. Training progress of the self-organizing map. Figure 2. Training progress of of the the self-organizing self-organizing map. map. Figure self-organizing map. Figure 2. 2. Training Training progress # plot how many cells are mapped to each neuron Figure 2. Training progress of the self-organizing map. # plot plot how how many many cells are are mapped mapped to each each neuron neuron plot(som.res, type="counts", palette.name=coolBlueHotRed) # cells to plot(som.res, type="counts", palette.name=coolBlueHotRed) plot(som.res, type="counts", palette.name=coolBlueHotRed) > # plot how many cells are mapped to each neuron

> # U-Matrix to explore clusters palette.name=coolBlueHotRed) > plot(som.res, type="counts", > # U-Matrix to explore clusters palette.name=coolBlueHotRed) > plot(som.res, # U-Matrix to type="dist.neighbours", explore clusters > # U-Matrix to explore clusters plot(som.res, type="dist.neighbours", palette.name=coolBlueHotRed) > # areas with large distances refer to boundaries between clusters plot(som.res, type="dist.neighbours", palette.name=coolBlueHotRed) > plot(som.res, type="dist.neighbours", palette.name=coolBlueHotRed) # areas with large distances refer to boundaries between clusters > colnames(som.res$code) # plots normalized component planes # areas with large distances refer to boundaries between clusters > # areas with large distances refer to boundaries between clusters > colnames(som.res$code) # plots normalized component planes [1] "NATI" "HOUS" "PROX" "DENS" "VARI" > colnames(som.res$code) # plots normalized component planes > colnames(som.res$code) # plots normalized component planes [1] "NATI" "HOUS" "PROX" "DENS" "VARI" "VARI" > plot(som.res, type = "property", property = som.res$codes[,"PROX"], [1] "NATI" "HOUS" "PROX" "DENS" [1] "NATI" "HOUS" "PROX" "DENS" "VARI" > main=c("Proximity"), plot(som.res, type type = = palette.name=coolBlueHotRed) "property", property property = som.res$codes[,"PROX"], som.res$codes[,"PROX"], + > plot(som.res, "property", > plot(som.res, type = "property", property == som.res$codes[,"PROX"], + main=c("Proximity"), palette.name=coolBlueHotRed) > plot(som.res, type = "property", property = som.res$codes[,"DENS"], + main=c("Proximity"), palette.name=coolBlueHotRed) + main=c("Proximity"), palette.name=coolBlueHotRed) > main=c("Density"), plot(som.res, type = "property", "property", property = som.res$codes[,"DENS"], som.res$codes[,"DENS"], + palette.name=coolBlueHotRed) > plot(som.res, type = som.res$codes[,"DENS"], > plot(som.res, type = "property", property property == + main=c("Density"), palette.name=coolBlueHotRed) + main=c("Density"), palette.name=coolBlueHotRed) + main=c("Density"), palette.name=coolBlueHotRed)

(a)

(a) (a) (a)

Figure 3. Cont.

Data 2017, 2, 7 Data 2017, 2017, 2, 2, 77 Data

8 of 10 88 of of 10 10

Data 2017, 2, 7

8 of 10

(b) (b) (b) Figure 3. Component planesfor forthe the variables variables (a) and (b)(b) density. Figure 3. Component planes (a)proximity proximity and density. Figure Figure 3. 3. Component Component planes planes for for the the variables variables (a) (a) proximity proximity and (b) density. In the next step, the grid of the SOM is clustered by means of the k-means algorithm. To analyze In the next step, the grid of the SOM is clustered by means of the k-means algorithm. To analyze an appropriate of of clusters, we is loop throughbybetween twothe and ten clusters, and for In the next step,number the grid the SOM clustered means of k-means algorithm. Toeach analyze In the next step, the grid of the SOM is clustered means oftwo the and k-means algorithm. Tofor analyze an appropriate number of clusters, we loop throughbybetween ten clusters, and each clustering, the within cluster sum of squares is computed. Plotting these values shows an elbow an appropriate number of clusters, we loop through between two and ten clusters, and forat each an appropriate number of clusters, we loop is through between twothese and ten clusters, and for each clustering, the within cluster sum of squares computed. Plotting values shows an elbow at seven clusters, referring to asum suitable solution.is computed. Plotting these values shows an elbow at clustering, the within cluster of squares

clustering, the within cluster sum of solution. squares is computed. Plotting these values shows an elbow at seven clusters, referring to a suitable seven clusters, referring to a suitable solution. seven>clusters, referring to a suitable # cluster component planes solution. through k-means and > > > > > > > > > > > > + + + + + + > >

> # determine numberplanes of clusters by means of minimizing the # cluster component through k-means and # cluster component planes through k-means and > # within cluster of sum of squares # determine number clusters by means of minimizing the # determine number of clusters by means of minimizing the > som.cluster wss > >

> som.cluster.res # plot the clustered component planes > add.cluster.boundaries(som.res, som.cluster.res) > # plot the the clustered component planes > # plot clustered component planes > plot(som.res, type="mapping", bgcol = brewer.pal(7, "Set1")[som.cluster.res], > plot(som.res, type="mapping", bgcol = brewer.pal(7, "Set1")[som.cluster.res], main >=plot(som.res, "Clusters") type="mapping", bgcol = brewer.pal(7, "Set1")[som.cluster.res], main main = "Clusters") = "Clusters") > add.cluster.boundaries(som.res, som.cluster.res) > add.cluster.boundaries(som.res, > add.cluster.boundaries(som.res,som.cluster.res) som.cluster.res) > # plot the clustered component planes > # plot the clustered component planes > plot(som.res, type="mapping", bgcol = brewer.pal(7, "Set1")[som.cluster.res], > # plot clusters on a geographical > plot(som.res, type="mapping", bgcolmap = brewer.pal(7, "Set1")[som.cluster.res], main >=cluster_details "Clusters") add.cluster.boundaries(som.res, som.cluster.res) cluster=som.cluster.res[som.res$unit.classif]) > add.cluster.boundaries(som.res, som.cluster.res) > # plot clusters on a geographical map > # plot clusters on a geographical map > cluster_details cluster_details data@data data@data data$fcluster data$fcluster spplot(data, "fcluster", col.regions=brewer.pal(7, "Set1")) > spplot(data, "fcluster",col="transparent", col="transparent", col.regions=brewer.pal(7, "Set1"))

Figure 4. Result of the SOM-based clustering. Figure 4. Result of the SOM-based clustering.

Exploring the descriptives of each cluster indicate that cluster 5 could be related to pockets of Figure 4. Result of the SOM-based clustering.

Exploring descriptives of each cluster indicate that cluster 5 could be related to pockets food deserts,the even though the corresponding cells are exclusively located in the urban periphery. This of food challenges deserts, even though the corresponding cellsbut areconfirms exclusively located in theaccessibility urban periphery. the classic interpretation of food deserts differences in spatial to Exploring the descriptives of each cluster indicate that cluster 5 could be related to pockets of healthy food supplied by supermarkets and area-based socioeconomic characteristics within the city This challenges the classic interpretation of food deserts but confirms differences in spatial accessibility food of deserts, even though the corresponding cells arewas exclusively located the in the urban periphery. This Amsterdam. In conclusion, no empirical evidence supportingcharacteristics notion of pronounced to healthy food supplied by supermarkets and area-basedfound socioeconomic within the city challenges the classic interpretation of food deserts but confirms differences in spatial accessibility to inequalities In in conclusion, access to healthy food. of Amsterdam. no empirical evidence was found supporting the notion of pronounced healthy food supplied by supermarkets and area-based socioeconomic characteristics within the city Acknowledgments: This study was supported by the interdisciplinary research program Healthy Urban Living inequalities in access to healthy food. of Amsterdam. In conclusion, no empirical evidence was found supporting the notion of pronounced of Utrecht University (www.uu.nl/hul). inequalities in access healthy Acknowledgments: Thistostudy wasfood. supported by the interdisciplinary research program Healthy Urban Living Author Contributions: MH developed the idea and the study design. MH and JH did the data analysis. MH

of Utrecht University (www.uu.nl/hul). wrote the first draft ofstudy the manuscript. Both authors and approved the final manuscript. Acknowledgments: This was supported by theread interdisciplinary research program Healthy Urban Living Author Contributions: M.H. developed the idea and the study design. M.H. and J.H. did the data analysis. of Utrecht University (www.uu.nl/hul). Conflicts of Interest: The authors declare no conflict of interest. M.H. wrote the first draft of the manuscript. Both authors read and approved the final manuscript. Author Contributions: MH developed the idea and the study design. MH and JH did the data analysis. MH Conflicts of Interest: The authors declare no conflict of interest. wroteReferences the first draft of the manuscript. Both authors read and approved the final manuscript. 1.

Beaulac, J.; Kristjansson, E.; Cummins, S. A systematic review of food deserts: 1966–2007. Prev. Chronic Dis. 2009, 6, 1–10. 2. Zenk, Schulz, A.J.; B.A.; James, S.A.; Bao, S.; Wilson, M.L. Neighborhood racial composition, 1. Beaulac, J.; S.N.; Kristjansson, E.; Israel, Cummins, S. A systematic review of food deserts: 1966–2007. Prev. Chronic Dis. References poverty, and the spatial accessibility of supermarkets in metropolitan Detorit. Am. J. Public 2009,neighborhood 6, 1–10. 95, 660–667. 1. Beaulac, J.; 2005, Kristjansson, E.; Cummins, S. A systematic review foodWilson, deserts: 1966–2007. Prev. Chronic Dis. 2. Zenk,Health S.N.; Schulz, A.J.; Israel, B.A.; James, S.A.; Bao,of S.; M.L. Neighborhood racial 3. Cummins, S.; Macintyre, S. “Food deserts”—Evidence and assumption in health policy making. Br. Med. J. 2009, 6, 1–10. neighborhood poverty, and the spatial accessibility of supermarkets in metropolitan Detorit. composition, 2002, 25, 436–438. 2. Zenk,J. S.N.; A.J.; Israel, B.A.; James, S.A.;[PubMed] Bao, S.; Wilson, M.L. Neighborhood racial composition, Am. PublicSchulz, Health 2005, 95, 660–667. [CrossRef] 4. Charreire, H.; Casey, R.; Salze, P.; Simon, C.; Chaix, B.; Banos, A.; Badariotti, D.; Weber, C.; Oppert, J.M. neighborhood poverty, and the spatial accessibility of and supermarkets in in metropolitan Detorit. Am. PublicJ. 3. Cummins, S.; Macintyre, S. “Food deserts”—Evidence assumption health policy making. Br.J.Med. Measuring the food environment using geographical information systems: A methodological review. Public Health 2005, 95, 660–667. 2002,Health 25, 436–438. [CrossRef] Nutr. 2010, 13, 1773–1785.

Conflicts of Interest: The authors declare no conflict of interest. References

3. 4. 4.

Cummins, S.; Macintyre, “Food and assumption in healthD.; policy making. Br. Med. J. Charreire, H.; Casey, R.; S. Salze, P.; deserts”—Evidence Simon, C.; Chaix, B.; Banos, A.; Badariotti, Weber, C.; Oppert, J.M. 2002, 25, 436–438. Measuring the food environment using geographical information systems: A methodological review. Charreire, H.;Nutr. Casey, R.;13, Salze, P.; Simon, C.; Chaix, B.; Banos, A.; Badariotti, D.; Weber, C.; Oppert, J.M. Public Health 2010, 1773–1785. [CrossRef] [PubMed] Measuring the food environment using geographical information systems: A methodological review. Public Health Nutr. 2010, 13, 1773–1785.

Data 2017, 2, 7

5. 6. 7. 8. 9. 10. 11.

12. 13. 14. 15. 16. 17. 18. 19. 20.

21. 22. 23. 24. 25.

26. 27. 28.

10 of 10

Apparicio, P.; Cloutier, M.; Shearmur, R. The case of Montreal’s missing food deserts: Evaluation of accessibility to food supermarkets. Int. J. Health Geogr. 2007, 6, 4. [CrossRef] [PubMed] Russell, S.; Heidkamp, C. “Food desertification”: The loss of a major supermarket in New Haven, Connecticut. Appl. Geogr. 2011, 31, 1197–1209. [CrossRef] Wang, H.; Qiu, F.; Swallow, B. Can community gardens and farmers’ markets relieve food desert problems? A study of Edmonton, Canada. Appl. Geogr. 2014, 55, 127–137. [CrossRef] Brunsdon, C. Quantitative methods I: Reproducible research and quantitative geography. Prog. Hum. Geogr. 2015, 1–10. [CrossRef] Helbich, M.; Schadenberg, B.; Hagenauer, J.; Poelman, M. Food Deserts? Healthy Food Access in Amsterdam. 2017, under revision. GGD Amsterdam. Amsterdammers Gezond en Wel? Available online: http://www.ggd.amsterdam.nl/ beleid-onderzoek/gezondheidsmonitors/amsterdamse/ (accessed on 2 October 2016). Centraal Bureau voor de Statistiek. Statistische Gegevens per Vierkant. Den Haag/Heerlen. Available online: http://www.cbs.nl/NR/rdonlyres/E29D852E-0AD5--40AD-AF57-C42F811487B6/0/ Statistischegegevenspervierkantupdateoktober2014.pdf (accessed on 2 October 2016). Wikipedia. Lijst van Supermarkten. Available online: https://nl.wikipedia.org/wiki/Lijst_van_supermarkten (accessed on 2 October 2016). Levensmiddelen Krant. Available online: http://www.levensmiddelenkrant.nl/ (accessed on 2 October 2016). BAG. Kadaster. Available online: https://www.kadaster.nl/bag (accessed on 2 October 2016). R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria. Available online: http://www.R-project.org/ (accessed on 2 October 2016). Hagenauer, J.; Helbich, M. Contextual neural gas for spatial clustering and analysis. Int. J. Geogr. Inf. Sci. 2013, 27, 251–266. [CrossRef] Hagenauer, J.; Helbich, M. SPAWNN: A toolkit for spatial analysis with self-organizing neural networks. Trans. GIS 2016, 20, 755–774. [CrossRef] Helbich, M.; Blüml, V.; Leitner, M.; Kapusta, N.; Stadsgeografie, S. Does altitude moderate the impact of lithium on suicide? The case of Austria. Geospat. Health 2013, 7, 209–218. [CrossRef] [PubMed] Kohonen, T. Self-Organizing Maps; Springer: New York, NY, USA, 2001. Helbich, M.; Hagenauer, J.; Leitner, M.; Edward, R. Exploration of unstructured narrative crime reports—An unsupervised neural network and point pattern analysis approach. Cartogr. Geogr. Inf. Sci. 2013, 40, 326–336. [CrossRef] Helbich, M.; Brunauer, W.; Hagenauer, J.; Leitner, M. Data-driven regionalization of housing markets. Ann. Assoc. Am. Geogr. 2013, 103, 871–889. [CrossRef] Kulldorff, M. A spatial scan statistic. Commun. Stat. Theory Methods 1997, 26, 1481–1496. [CrossRef] Helbich, M. Beyond postsuburbia? Multifunctional service agglomeration in Vienna’s urban fringe. Tijdschrift voor Economische en Sociale Geografie 2012, 103, 39–52. [CrossRef] Pebesma, E.; Bivand, R. Classes and methods for spatial data in R. R News 2005, 5. Available online: http://cran.r-project.org/doc/Rnews/ (accessed on 10 October 2016). Bivand, R.; Keitt, T.; Rowlingson, B. Rgdal: Bindings for the Geospatial Data Abstraction Library. R Package Version 1.1-10. 2016. Available online: https://CRAN.R-project.org/package=rgdal (accessed on 10 October 2016). Bivand, R.; Piras, G. Comparing implementations of estimation methods for spatial econometrics. J. Stat. Softw. 2015, 63, 1–36. [CrossRef] Wehrens, R.; Buydens, L. Self- and super-organising maps in R: The kohonen package. J. Stat. Softw. 2007, 21, 1–19. [CrossRef] Neuwirth, E. RColorBrewer: ColorBrewer Palettes. R Package Version 1.1-2. 2014. Available online: https://CRAN.R-project.org/package=RColorBrewer (accessed on 10 October 2016). © 2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).