ArticlePDF Available

Abstract and Figures

This study revisits the debate surrounding the definition of neighborhood boundaries by addressing the disconnect between the city’s Administrative Neighborhoods and its functional organization. A method is proposed for dividing the city into more meaningful units through the spatial distribution of urban activities by retrieving data from Google Places. The dataset was pre-processed and spatially divided into Functional Clusters. A comparison between functional and administrative subdivisions of the city was undertaken, from which three overall conclusions could be drawn. First, a function-based city partition allows economically active urban areas to become the neighborhood’s center, thereby creating a polynuclear neighborhood structure that would potentially encourage greater cross-movement of people throughout the city. Second, the specialization of activities becomes more evident in Functional Clusters than in Administrative Neighborhoods. Third, access to up-to-date data makes possible a timely diagnosis of the quantity and diversity of urban activities—i.e., economic activities, services, and facilities— through Google Places data. The value of this contribution is to inform urban decision-making and policies in order to better balance the provision of a neighborhood’s economic activity.
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
Journal of Urban Technology
ISSN: (Print) (Online) Journal homepage:
Revisiting the Spatial Definition of Neighborhood
Boundaries: Functional Clusters versus
Administrative Neighborhoods
Pablo Martí, Leticia Serrano-Estrada, Almudena Nolasco-Cirugeda & Jesús
López Baeza
To cite this article: Pablo Martí, Leticia Serrano-Estrada, Almudena Nolasco-Cirugeda &
Jesús López Baeza (2021): Revisiting the Spatial Definition of Neighborhood Boundaries:
Functional Clusters versus Administrative Neighborhoods, Journal of Urban Technology, DOI:
To link to this article:
© 2021 The Author(s). Published by Informa
UK Limited, trading as Taylor & Francis
View supplementary material
Published online: 01 Jul 2021. Submit your article to this journal
Article views: 372 View related articles
View Crossmark data
Revisiting the Spatial Denition of Neighborhood Boundaries:
Functional Clusters versus Administrative Neighborhoods
Pablo Martí
, Leticia Serrano-Estrada
, Almudena Nolasco-Cirugeda
, and
Jesús López Baeza
Building Sciences and Urbanism Department, University of Alicante, Alicante, Spain;
Digital City Science,
HafenCity University Hamburg, Germany
This study revisits the debate surrounding the denition of
neighborhood boundaries by addressing the disconnect between
the citys Administrative Neighborhoods and its functional
organization. A method is proposed for dividing the city into
more meaningful units through the spatial distribution of urban
activities by retrieving data from Google Places. The dataset was
pre-processed and spatially divided into Functional Clusters. A
comparison between functional and administrative subdivisions
of the city was undertaken, from which three overall conclusions
could be drawn. First, a function-based city partition allows
economically active urban areas to become the neighborhoods
center, thereby creating a polynuclear neighborhood structure
that would potentially encourage greater cross-movement of
people throughout the city. Second, the specialization of
activities becomes more evident in Functional Clusters than in
Administrative Neighborhoods. Third, access to up-to-date data
makes possible a timely diagnosis of the quantity and diversity of
urban activitiesi.e., economic activities, services, and facilities
through Google Places data. The value of this contribution is to
inform urban decision-making and policies in order to better
balance the provision of a neighborhoods economic activity.
Neighborhood boundaries;
functional clusters; urban
economic activities; Google
Places; social networks
Economic activity growth in cities has brought new challenges to the design of urban inter-
vention strategies (Kärrholm, 2012). As suggested by Zukin (1993), the transformation of
city landscapes of productioninto landscapes of consumptionhas aected urban plan-
ning and the design of city spaces (Kärrholm, 2012; Mubi Brighenti and Kärrholm, 2018).
Moreover, the morpho-spatial reconguration of cities is greatly inuenced by the concen-
tration of economic activities (Kärrholm, 2012;SaraivaandPinho,2016). Indeed, the link
between the physical aspects of the urban form and the clustering of city functions contrib-
utes to create a specic type of urban life (Crooks et al., 2015;Kropf,1996).
© 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License
(, which permits non-commercial re-use, distribution, and reproduction in any
medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
CONTACT Dr Pablo Martí University of Alicante, Carretera San Vicente del Raspeig s/n, 03690
San Vicente del Raspeig, Alicante, Spain.
This article has been corrected with minor changes. These changes do not impact the academic content of the article.
Urban activity together with other factors (i.e., social preferences and urban mor-
phology) have a signicant impact on the perceived character of neighborhoods (Cran-
shaw et al., 2012; Kropf, 1996). In fact, the availability and easy access to the range of
goods and services that provide for residentsnecessities and amenities give neighbor-
hoods a sense of place and identity (Mehta and Mahato, 2018:2). Indeed, at cognitive
and perceptual levels, neighborhood units are largely dened by the association of
spatial features to socioeconomic and behavioral aspects (Kropf, 1996; Lynch, 1960).
On this basis, many authors have highlighted the importance of considering the types
of functions and establishments when partitioning the city (Brown, 1991,1993; Cran-
shaw et al., 2012; Rösler and Liebig, 2013; Saraiva and Pinho, 2016) given that a concen-
tration of specic economic urban environments may contribute to the fragmentation of
the urban landscape (Graham and Marvin, 2002).
Traditionally, the catchment area of specic servicestypically dened by administra-
tive partitionshas been a key criterion for the denition of neighborhood boundaries
(Devine, 1975; Perry, 1929). Frequently, these administrative partitions do not correspond
to the current functional organization of the city. Lack of awareness of this disconnect chal-
lenges the eective provision of economic activities, services, and facilities, aecting social
indicators such as equity and accessibility planning (Abdullahi et al., 2018;Omar,2019;
Serag El Din et al., 2013), which may thereby result in poor decision-making while
dening urban intervention and regeneration strategies. The identication of neighbor-
hood units regarded as functional urban areas has been approached from a variety of per-
spectives in the Urban Studies and Planning literature (Chen et al., 2017). Research
dedicated to this issue contributes towards a better understanding of the character of the
urban landscape and makes some headway in dening accurate public policies and enhan-
cing the development of more eective urban intervention proposals (Gu, 2010; Kropf,
1996; Saraiva and Pinho, 2016). Grounded in the abovementioned theoretical reasoning,
this research addresses the knowledge gap identied by Ellen and Turner (1997:855)
regarding the need for a method to dene more meaningful neighborhood boundaries.
Therefore, the aim of this study is to propose a method for partitioning the city into mean-
ingful neighborhood unitshereafter referred to as Functional Clustersthat reect the
spatial distribution of economic activities, services, and facilities. The proposed method
is applied to an illustrative case study, where the delimitation of the Administrative Neigh-
borhoods is compared to the Functional Clusters. The hypothesis is that Functional Clus-
ters better reect the character of the urban environment than existing traditional
Administrative Neighborhood partitions.
The novelty of this research is twofold. First, the method for obtaining Functional
Clusters uses Google Places data as a source of information on todays urban activity
and elaborates a process that allows dierent degrees of granularity in the analysis.
Second, the comparison between Functional Clusters and Administrative Neighbor-
hoods contributes to the contemporary methodological and theoretical debate on the
spatial denition of neighborhood boundaries.
This paper is structured as follows: the literature section covers some of the most rel-
evant research on both the denition of spatial neighborhood boundaries and the assess-
ment of spatial patterns related to urban activity through social media; next, a description
of the illustrative case study and the sources is given, followed by an explanation of the
method; nally, the results are presented, discussed, and the main conclusions are drawn.
Literature Review
Spatial Neighborhood Boundaries
Jenks and Dempsey (2007) provided a comprehensive outline of the various scholarly
denitions of the term neighborhood,but, in general terms, scholars agree that neigh-
borhoods are a geographically dened subarea of the city where residents are presumed
to share both spatial proximity and some degree of mutual circumstance(Chaskin et al.,
2001:98). Furthermore, these are subareas of towns and cities whose physical or social
characteristics distinguish them from one another(Rohe, 2009:99). Thus, the term
neighborhoodencompasses spatial and non-spatial features, such as, intangible
social elements (Jenks and Dempsey, 2007). Therefore, when both place and human
activity are interlinkedas Hallman (1984:102) suggested, through occupancy and
use by its residents”—a geographical area can be considered a neighborhood.
Nonetheless, the delimitation of the citys spatial subdivisions based merely on human
activity is a challenging task, which is one of the main reasons for the controversial and
longstanding debate on the relevant determinants for dening neighborhood boundaries
(Jenks and Dempsey, 2007). Previous work in the eld has considered city subdivisions
based on census tracts and administrative areas (Steiger et al., 2015), statistically dened
clusters (Martin, 1998), and land use classication (García-Palomares et al., 2017). As
these numerically-based subdivision criteria may fail to accurately represent the neigh-
borhood conditions that make a dierence in peoples lives(Ellen and Turner,
1997:844), other more citizen-oriented approaches have considered perceptual and cog-
nitive information for identifying neighborhood boundaries (Cranshaw et al., 2012;
Nejat, 2018).
This paper adopts Hallmansdenition of neighborhoods as functional entities that
support residentsneeds for services and facilities (Hallman, 1984; Jenks and Dempsey,
2007), taking into account that the quantity, quality, and diversity of uses and functions
in a neighborhood encourage or discourage activities that involve using and staying in
urban spaces (Rohe, 2009). These factors aect social and economic dimensions (Ellen
and Turner, 1997) and, in turn, the urban livability of the space (Mehta and Mahato,
2018). This approach recognizes that communities can be weakened by the dispersion
of people and activities (The Urban Task Force, 2003) or the lack of nearby facilities, insti-
tutions, and services, which forces residents to concentrate activity and connections
beyond rather than within the neighborhoodspatial limits (Chaskin et al., 2001:9; Fur-
stenberg, 1993). Therefore, dening more meaningful boundaries based on the availability
and proximity of urban activities could potentially encourage residents to stay within the
neighborhood, thus supporting the livability of its urban spaces.
Urban Activity Patterns through Location-Based Social Networks
Pioneering research in the identication of geographical urban activity patternsspatial
distribution of economic activities, services, and facilitieswas traditionally carried out
by using store-by-store and street-by-street data collection and clustering techniques
(Guy, 1976; Lee and McCracken, 1982; Saraiva and Pinho, 2016). The conclusions of
this approach are similar to those that use dierent techniques involving more techno-
logical processes for data collection and categorization, as well as spatial analysis and
clustering (Hossain, 1999; Saraiva, 2013; Sarma, 2006). In situ data collection is comple-
mented and/or substituted by online catalogues, or ocial registers from local govern-
ments and other institutions, such as National Statistics Institute, commerce
associations, etc. Clustering, among other types of analysis, has been calculated by
using specialized software.
Nowadays, the great data collecting potential of crowd-sourced social media is being
exploited to analyze a diverse range of topics related to the functional organization of the
city (Arribas-Bel and Tranos, 2018), such as the relationship between urban form and
function (Crooks, Pfoser et al., 2015; Crooks, Croitoru et al., 2016); the identication
of POIspoints of interest(Van Canneyt et al., 2012; Deng and Newsam, 2017;
García-Palomares et al., 2015; Van Weerdenburg et al., 2019) and their accessibility in
terms of density and diversity (Shen and Karimi, 2016); the characterization of liveli-
hoods according to the collective behaviors of residents (Cranshaw et al., 2012); and
the delimitation of functional areas to understand social and spatio-temporal aspects
of the city (Chen et al., 2017; Rösler and Liebig, 2013).
In line with these studies, this research sources geolocated data from the social
network Google Places, which has previously been used to identify economic activity
for measuring urban entropy and urban complexity (Bustos Hernándes, 2015; López
Baeza et al., 2017; Serrano-Estrada et al., 2016), as well as to research socioeconomic pat-
terns of urban areas (Cenamor et al., 2017).
Illustrative Case Study and Sources
Alicante City, Spain was used as an illustrative case study as it is representative of other
mid-sized Mediterranean cities that have encouraged modern commercial, tourist, and
leisure formats in the last decades (Nieto, 2015). This provincial capital, at the heart of
the Spanish Mediterranean Arc, with 329,988 registered inhabitants (Spanish Statistics
Institute, 2018), is considered an important European holiday destination hub (Suau-
Sanchez and Burghouwt, 2012). Moreover, like other Mediterranean cities, Alicante
has at least two dierent types of administrative spatial divisions that do not correspond
with each other, nor do they reect the social or functional reality of the city. For
instance, this municipality is divided into eight census districts, which are subdivided
into 253 census sections (Spanish Statistics Institute, 2018), and into 42 Administrative
Neighborhood areas according to the ocial database of the Urban Guide of the City of
Alicante (Ayuntamiento de Alicante, 2018). The latter division reects the historical
growth of the city, observable in the existing patterns of the urban fabric. However,
neither division considers todays dynamic socioeconomic perspective suggested by,
for example, the distribution and types of urban activities (Noulas et al., 2013) that are
crucial for determining the direction of urban interventions (Mankalpa et al., 2015;
Temes Cordóvez et al., 2016).
Two main sources were used for this study: The administrative neighborhood spatial
delimitation of Alicante municipality and the geo-location data from Google Places. The
spatial delimitation of Alicantes 42 administrative neighborhoods (Ayuntamiento de
Alicante, 2018) was used as the citys baseline division. The spatial extent of all admin-
istrative neighborhoods together was considered as the area within which Google
Places data were collected. The software program Social Media Urban Analyzer
(SMUA) (Martí et al., 2019), which retrieves data via Google Places was used for this
purpose. The data from Google Places were retrieved as a listing of placeswhich
include all urban activities and open public spaces registered in the platform within
the search area up to the date of retrievalJanuary 23, 2018. The raw dataset included,
in total, 57,578 geo-referenced places.
Each listed place has associated metadata. From the large quantity of data variables
available through the Google Places API, this research selects and retrieves only the fol-
lowing: the spatial coordinateslatitude and longitudethe name of the place; the ID
place unique identier numberand the place type, which are tags or descriptors.
These place types in Google Places datasets are relevant for this study since they
provide detailed information as to what kind of urban activity occurs in each place
Identifying and Delimiting Functional Clusters Using Google Places Data
The method adopted consisted of three phases. First, data pre-processing was necessary
to rene the places in the Google Places dataset. Second, the place types were categorized
into the hierarchical APA benchmark categories (American Planning Association,
2018a). Finally, identication, spatial delimitation, and analysis of Functional Clusters
were carried out.
Pre-Processing of Google Places Retrieved Data
The pre-processing of the raw dataset involved the following stages (See Figure 1):
Figure 1. Pre-processing data summary
.elimination of duplicate data
.validation of places related to urban activitiesspecically economic activities, ser-
vices, and facilities
.re-assignment of Google Places place types when necessary.
These steps ensured that each datapoint represented a unique register and was correctly
assigned to a pre-dened Google Places place type (Google Developers, 2018).
Elimination of Duplicate Place Listings. A script was designed to eliminate duplicate
place listings by considering three specic data variables: name of the place; place ID;
and geographic coordinates.
Google Places assigns a unique ID to each place (Google Developers, 2018), and the
criteria adopted for eliminating duplicate listings were twofold. Places are considered
duplicate listings if they have the same ID, or if they have a dierent ID but the same
name and the same geographic coordinates. Those places with dierent IDs, dierent
names, and the same geographic coordinates are considered unique as they may, for
example, be dierent economic activities within the same building.
Validation of Places. Recognizing which datapoints represent economic activities, ser-
vices, and facilities entailed revising their originally assigned place types. In total,
Google Places currently supports 128 standard predened place types. However, the
raw retrieved dataset for Alicante included only 103 place types. This is because the diver-
sity of urban activities changes from one city to another.
Another relevant consideration is that place types are automatically assigned by
Google or manually designated by the individual who is registering a service or facility
in the platform. Thus, according to the authorsexperience, some places may include
more or fewer place types or descriptors than others.
Some of these place types are ambiguous and misleading; for example, the establish-
ment place type which, as opposed to electronics_store, does not oer enough information
on the kind of economic activity that takes place in the space. In view of the above obser-
vations, this study adopts the most specic place type assigned to each place.
After eliminating duplicate registers, three types of datapoints were recognizable: valid
datapoints, invalid datapoints, and checkable datapoints. The valid datapoints are those
places which represent mostly urban activities, i.e., economic activities, services, and
facilities. These include a signicant number of datapoints assigned to the generic
store place type which does not provide enough information about the type of activity.
The registers tagged with this ambiguous place type have been checked and reassigned,
where possible, a more specic place type.
The invalid datapoints are those places that do not represent services or facilities and,
therefore, were excluded. Invalid datapoints included in the case study dataset can be
grouped as follows:
.geometric regions that do not represent services or facilities: administrative_area_le-
vel_1, 2, 3, 4, and 5; colloquial_area; country; geocode; locality; neighborhood; political;
postal_ code; postal_code_prex; postal_code_sux; postal_town; sublocality; subloca-
lity_level_1, 2, 3, 4, and 5; and subpremise.
.datapoints that represent street names and/or are assigned by Google Places to the
place types: intersection; route; street_address; and street_number.
.datapoints that represent places with no economic activity. These are mostly the place
types: parks; plazas; natural_features; landmarks; and campground.
The checkable datapoints that require reviewing before being considered valid or invalid
datapoints can be divided into two groups:
(1) Datapoints assigned to the lodging place type include dierent types of lodgings
(apartments, residential buildings, hotels, etc.) that, in some cases, may be con-
sidered as an economic activity, for example, hotels. Residential only buildings are
not considered as an economic activity.
(2) Datapoints assigned to several broad and generic place types, namely, Establishment,
Point_of_Interest, and Premise (EPP), that may or may not contain economic activi-
ties and would be better assigned to a more specic place type that reects their type
of activity.
Re-Assigning Place Types to Google Places Checkable Datapoints. This step was useful
for the grouping and homogenization of data into rened place types. This procedure
consisted of rst, assigning datapoints from the categories EPP and store to a specic
pre-dened Google Places place type; and second, separating and assigning the hotel
place type to hotels that were originally tagged as lodging and not as hotel.
From all the EPP datapoints retrieved (10,669), only six were originally assigned to both
premise and park place types. As for the rest, it was necessary to resort to the place name vari-
able to identify the type of activity of each datapoint and reassign a place type. All the words
included in the EPP datapoint place names weresortedbyfrequency,usingasimplePython
and the 1,500 most frequented terms were closely analyzed to see if they could be
associated with a pre-dened Google Places place type. For instance, place names containing
the words restaurant,”“meson,and casa de comidaswere assigned to the Google Places
food place type. Terms related to street names were considered invalid EPP datapoints.
Finally, 76 places were eliminated as they did not represent an economic activity.
The store place type was too generic and thus, the same process was adopted for reas-
signing place types. Datapoints originally assigned as lodging were reassigned a place
type according to their kind of accommodation. Only those datapoints reassigned to
the hotel category have been considered as economic activities:
.Residential: residential buildings and condosurbanización, edicio residencial
.Hotel: tourist accommodation, such as hotels and hostels
.Apartment: places dened as rental apartmentsshort-term accommodation
.Lodging: places that could not be included in the previously mentioned place types.
Some of these places have been registered in Google Places with the name of a
person or a random name that does not provide further information about the type
of accommodation (e.g., La Casita, La Lonja, and Maralic).
Additionally, new place types were created for those places whose names provided
enough information about the type of economic activity, but no suitable predened
Google Places place type was found. These new place types are: association; cleaning_ser-
vice; design_studio; events_venue; industry; leisure; logistics; lottery; professional_service;
theater; and, electronics_repair. Over 25.94 percent of EPP datapoints were assigned a
place type using the described criteria and 76 (9.20 percent) datapoints in the lodging
category were assigned to the hotel place type.
Rened Google Places Dataset
After data pre-processing, there were 23,845 operating datapoints remaining in the
rened Google Places dataset (See Figure 1) which represent 70 percent of the unique
dataset datapoints. These included: 21,002 originally valid datapoints; 2,767 EPP data-
points with an assigned Google Places place type; and 76 hotel datapoints. As for the
number of rened place types, considering both Google Places and newly created place
types, 107 place types were obtained in total.
Recategorizing Operating Datapoints: from Google Places Categories to APA
Land Based Classication Standards Benchmark Categories
In order to permit an analysis of the types of economic activities, services, and facilities
with dierent degrees of granularity, a grouping of the Google Places place types is pro-
posed following previous research from Martí et al. (2019). To this end, the Land Based
Classication Standards categories have been adopted (American Planning Association,
2018a) as benchmark categories. Specically, this classication is the one that addresses
the functional dimension,one of the ve available LBCS dimensions, which refers to
the economic function or type of establishment using the land(American Planning
Association, 2018b). This hierarchical classication provides an overall ne-grain land
use class taxonomy of nine Level 1 categories, 47 Level 2 categories, and 159 Level 3 cat-
egories (Deng and Newsam, 2017). All rened Google Places place types were assigned to
APA Level 1 (See Figure 2) and Level 2 (See Table 1) category codes, respectively.
From the 47 APA Level 2 categories originally available, the case studysGoogle Places
and newly proposed place types could only be assigned to 28 APA Level 2 categories. The
remaining 19 Level 2 categories have not been used. Specically, no place types were
assigned to the following APA Level 2 subcategories:
1100 Private household
1200 Housing services for the elderly
3200 Wood, paper, and printing products
3300 Chemicals, and metals, machinery, and electronics manufacturing
4300 Utilities and utility services
5500 Natural and other recreational parks
7200 Machinery related
7400 Heavy construction
8100 Oil and natural gas
8200 Metalsiron, copper, etc.
Table 1. Number of Google Places and newly proposed place types included in APA Level 2
subcategories according to the case studys operating datapoints
APA Level 2 sub-categories
Google Places place types and newly proposed place
Number of place
types per APA Level 2
1 1300 Hotels, motels, or other
accommodation services
hotel 1
2 2100 Retail sales or service store; orist; health; grocery_or_supermarket;
clothing_store; car_repair; gas_station; pharmacy;
locksmith; car_dealer; furniture_store; bicycle_store;
home_goods_store; book_store; car_wash;
electronics_store; shopping_mall; hardware_store;
shoe_store; liquor_store; convenience_store;
art_gallery; supermarket; department_store
3 2200 Finance and Insurance atm; nance; insurance_agency; bank 4
4 2300 Real estate, and rental and
real_estate_agency; car_rental 2
5 2400 Business, professional,
scientic, and technical
lawyer; professional_service; travel_agency;
copy_store; veterinary_care; accounting;
moving_company; design_studio; electronics_repair;
consulting; cleaning_service
6 2500 Food services food; bar; café; restaurant; meal_delivery; night_club;
7 2600 Personal services parlor; laundry; hair_care; beauty_salon 4
8 2700 Pet and animal sales or
service (except veterinary)
pet_store 1
9 3100 Food, textiles, and related
bakery 1
10 3400 Miscellaneous
industry; jewelry_store 1 1
11 3500 Wholesale trade
movie_rental 1
12 3600 Warehouse and storage
storage 1
13 4100 Transportation services parking; bus_station; light_rail_station; post_oce;
airport; taxi_stand; transit_station; train_station;
14 4200 Communications and
library 1
15 5100 Performing arts or
supporting establishment
stadium; movie theater; theater; events_venue 2 2
16 5200 Museums and other special
purpose recreational
museum; natural_feature; zoo 3
17 5300 Amusement, sports, or
recreation establishment
casino; gym; spa; amusement_park; bowling_alley;
leisure; lottery
18 5400 Camps, camping, and
related establishments
rv_park 1
19 6100 Educational services school; university 2
20 6200 Public administration courthouse; local_government_oce; city_hall 3
21 6300 Other government
embassy 1
22 6400 Public Safety re_station; police 2
23 6500 Health and human services doctor; physiotherapist; hospital; dentist 4
24 6600 Religious institutions hindu_temple; place_of_worship; mosque; church 4
25 6700 Death care services funeral_home; cemetery 2
26 6800 Associations, nonprot
organizations, etc.
association 1
27 7100 Building, developing, and
general contracting
general_contractor 1
28 7300 Special trade contractor electrician; roong_contractor; plumber; painter 4
8300 Coal
8400 Non-metallic mining
8500 Quarrying and stone cutting establishment
9100 Crop production
9200 Support functions for agriculture
9300 Animal production including slaughter
9400 Forestry and logging
9500 Fishing, hunting and trapping, game preserves
9900 Unclassiable function.
As can be observed, the majority of these subcategories correspond to the APA Level 1
categories (8000 Mining and extraction establishments, and 9000 Agriculture, forestry,
shing, and hunting) (See Figure 2).
The rened place types classication provides ne grain information as to specicusesin
an urban area. For instance, hair_care and beauty_salon are considered, in the rened
dataset,as two dierent placetypes. As previously stated, theAPA category classicationpro-
vides a hierarchical approach, grouping economic activities ofthe same type. In this case, the
hair_care and beauty_salon Google Places place types are grouped into the APA 2000 General
Sales and Services Level 1 (main category), 2600 Personal Services Level 2 (subcategory).
Therefore, in this research, the rened Google Places place types are used to infer the
diversity of economic activities, services, and facilities within each urban unit, whereas
the APA categorization is useful to recognize the functional character of clusters separ-
ately and in relation to the rest of the clusters.
Delimitation of Functional Clusters
The operating datapoints are visualized by means of cartography and grouped into 42
Functional Clusters (FC 0 to FC 41) by using Quantum GIS open-software. The
spatial delimitation of these 42 clusters is compared to the existing 42 Administrative
Neighborhoods (AN 1 to AN 42). This process comprised two phases: the cluster calcu-
lation, and the cluster borders denition.
A spatial cluster calculation by proximity of the rened datapoints was performed
using a k-Means clustering algorithm, a popular clustering method in literature,
adopted across dierent domains (Fortunato, 2010; Khan and Ahmad, 2004), and
Figure 2. Google Places operating datapoints and their respective place types assigned to APA Level 1
useful to partition topological nodes into groups. This algorithm presents quick conver-
gence to an acceptable result and, after a certain number of iterations, the position of the
centroids remains stable (Fortunato, 2010). The only problems raised in the literature are
the need for (a) experimenting with dierent numbers of clusters, and (b) close visual
exploration of the resulting clusters for further validation (Béjar et al., 2016). To over-
come these limitations several trial runs were conducted to test clustering results
yielded by dierent K-values (See Figure 3, left). The experiments were performed con-
sidering K-values from k=5 to k=50. It was observed that low K-values (i.e., k=5 and k=6)
resulted in rather dierent clustering results. However, higher K-values (i.e., k=40 to
k=51) yielded very similar clustering partitions (See Figure 3, right). Therefore, in this
specic case study, k=42 is selected as the K-value because the resulting clusters
provide consistencyas results started to converge around k=40, and it coincides with
the number of Administrative Neighborhoods, thus allowing a straightforward
Once the clusters were iteratively calculated by the k-Means algorithm and all data-
points were spatially linked to the centroid of the closest cluster using the Euclidean dis-
tance, each of the 42 clusters and their corresponding datapoints were given a code
number and a color to aid visual representation.
The Euclidean distance was considered instead of Manhattan or Network distances for
several reasons. First, no meaningful dierences were identied when compared to the
cluster subdivision using Manhattan distance, and second, the Euclidean distance is
quicker for computer calculations. Furthermore, the dierences between Manhattan
and Network distances are not signicant in the specic case of Alicante and are unlikely
to be in other Spanish Mediterranean cities whose urban fabric is mainly based on a
regular orthogonal grid (See Figure 4a and b).
The 42 clusters of datapoints obtained lack border lines. Therefore, a Voronoi diagram
was constructed for partitioning of the area into 23,845 coded Voronoi polygonal regions
Figure 3. Left: Voronoi experiments in increasing number of K-values from 5 to 42. Right: Superim-
posed representation of Voronoi experiments with high K-values ranging from 40 to 51
Figure 4. (a) Alicantes identied 42 Functional Clusters (b) Alicantes existing 42 Administrative
Neighborhoods; (c) Voronoi polygonal regions and (d) denition of Functional Clusters by grouping
these polygons; (e) Alicantes central area exemplifying how economically active segments of axes
fall within Functional Cluster areas
whose center is each datapoint (See Figure 4c). The aggregation of the coded Voronoi
polygonal regions forms shapes that ultimately dene the cluster borderlines (See
Figure 4d). Each spatial partition represents a grouped unit of economic activities, ser-
vices, and facilities, and thereby a Functional Cluster.
Administrative Neighborhoods vs. Delimited Functional Clusters
The Functional Clusters obtained were analyzed and compared to Administrative Neigh-
borhoods taking into account two dierent aspects: their physical characteristics (shape
and size), and, their functional features, in terms of the diversity and specialization of
economic activities, services, and facilities. For this purpose, the two data categorizations
dealt with in the previous sections were considered: the rened Google Places categories
and the benchmark APA categories.
Physical Characteristics
The resulting tessellation of Functional Clusters diers signicantly to that of the citys
recognized Administrative Neighborhoods. In fact, the size and shape of the polygon
delimitations are more homogeneous than those of the Administrative Neighborhoods.
Moreover, these polygons progressively increase in size as they get further away from the
city center because both the amount and the proximity of economic activities, services,
and facilities decreases in the less compact urban areas of the periphery.
Unlike what is observed in the Administrative Neighborhoods, where most active
street segments and intersections tend to be located along the neighborhood boundary
assigning each side of the street to dierent neighborhoods, in the Functional Cluster,
tessellation all of these active axes and nodes become the clusters core, thus creating a
polynuclear structure of city neighborhoods (See Figure 4e). Moreover, when one of
these active axes, due to its length, traverses two or more Functional Clusters, it is
observed that, in all cases, both sides of the street belong to the same Functional
Cluster. In general terms, a similar situation can be found in relation to intersections
and squares.
Exceptionally, for the city of Alicante, only one economically active intersection was
found to be located at the junction of two clusters. This is the case of the area occupied by
Plaza Luceros (See Figure 4e), a circle recognized as the most socially relevant open
public space of the city (Martí et al., 2017) in which the southwest area belongs to the
FC 39 while the rest of the square is connected to FC 19. Therefore, the space has two
functionallyand perceptiblydierent sides, each one strongly linked to nearby clus-
ters of economic activities, services, and facilities. This fact suggests that there could be
dierent signicance between the city scale and the cluster scale. Therefore, including
perceptual parameters of public spaces in the study of Functional Clusters could be an
interesting approach for future research.
Overall, the Functional Cluster tessellation obtained redenes axes according to their
actual function as an inter-neighborhood conduit of social life, rather than a neighbor-
hood boundary. Sometimes, the concentration of economic activities, services, and facili-
ties becomes a livable intersection, or sometimes it runs along an axis. In the rst case, the
aggregation of activities at the Functional Clusters center becomes the active
neighborhood nucleus, especially where these activities and services are located on
ground oors, as shown in Figure 4e), where the rened Google Places economic activities
categorized as APA 2100 and APA 2500 can be seen.
Functional Features
The analysis indicated that the spatial distribution of activity types is more homogeneous
in the 42 Functional Clusters than in the Administrative Neighborhoods.
In general terms, the economic activities of the case study fall within two predominant
Level 1 APA categories: category 2000 (general sales or services) and category 6000 (edu-
cation, public administration, health care and other institutions). In all 42 Functional
Clusters both categories were ranked rst and second respectively, based on their fre-
quency of economic activities.
A greater degree of granularity was obtained in the analysis as two approaches were
adopted to analyze and compare the functional features of both the Administrative
Neighborhoods and the Functional Clusters. On the one hand, the functional diversity
and the quantity of economic activities, services, and facilities of each cluster with
respect to the rest of the clusters were analyzed using the rened dataset; and, on the
other hand, the functional specialization of each cluster was identied by using APA
second-level categories. In both cases, the total number of economic activities, services,
and facilities per category was adopted as the main criteria.
In terms of the functional diversity, Figure 5 shows the number of Google Places place
types in both Functional Clusters and Administrative Neighborhoods. By comparing the
functional diversity of economic activities, services, and facilities between the Functional
Clusters and the Administrative Neighborhoods, it became apparent that the types of
activities were more evenly distributed in the former tessellation than in that of the
latter (See Figure 5). Clusters FC 40 and FC 5 are the two most relevant Functional Clus-
ters in terms of functional diversity with the presence of 25 dierent place types. While FC
5 is in a central location of the city, FC 40 is located in a residential tourist area that is very
well connected to other neighborhoods, and even municipalities, through main structural
axes. This cluster is characterized by the presence of large residential blocks, isolated high-
rise buildings, and single-family housing units, with mostly private and semi-public open
spaces. The economic activity, services, and facilities in cluster FC 40 run along the inter-
section of two axes, and are concentrated mainly in the ground oor of buildings. This
intersection happens to fall right at the geometric center of the Functional Cluster.
As for the quantity of economic activity, services, and facilities per category and per
cluster, Figure 5 shows that two out of the three most relevant clustersFC 3 and FC
15are located in the city center. The cluster ranked third is cluster FC 40 which, as
mentioned before, is also the most diverse. The ndings suggest that the diversity and
quantity of urban activities reects the functional character of the urban area in which
they are located. The city center has more quantity but a slightly less diverse economic
oer, being mostly stores, restaurants, bars, clothing stores, whereas residential areas
that are further away from the city center have more varied economic activity, services,
and facilities to satisfy the demand of local residents.
As for the clustersfunctional specialization, the presence of the APA second-level cat-
egories in each cluster can be observed in Figure 5. The diagram permits a twofold
reading: rst, the total number of places per type found in each cluster and, second, the
distribution of each type of economic activity across all clusters and neighborhoods. Both
the Administrative Neighborhoods and the Functional Clusters are compared. In the
former, there is a high concentration of activities (mainly in neighborhoods AN 5 and
AN 4) while in the latter, the spatial distribution of urban activity across all clusters is
more homogeneous.
Figure 5. Comparison between the diversity and quantity of APA second level categories and number
of Google Places place types within Administrative Neighborhoods (upper) and Functional Clusters
(lower). In parenthesis, the total number of place types (diversity) per neighborhood and cluster,
Furthermore, almost all Functional Clusters have a noticeable presence of two APA
second-level categories: 2100 Retail sales or services and 2500 Food Services. Thus, the
functional specialization becomes more evident with the proposed city subdivision
into Functional Clusters. Specically, worth highlighting is cluster FC 22, located at
the heart of the historic city center where not only does the predominant urban activity
correspond to businesses within the APA second-level category 2500 Food Services, but it
is also where most of these types of activities occur when compared to other Functional
Clusters. Moreover, clusters FC 8 and FC 40, two of the ve top ranking clusters in terms
of the amount of economic activity, services, and facilities present, are the most relevant
of all Functional Clusters in the transportation category. This may be a potential indi-
cator of their strong connectivity to other more central neighborhoods, while being func-
tionally independent.
Discussion and Conclusion
This research conrms that there is a signicant degree of disconnect between Alicantes
traditional Administrative Neighborhood partitions and the citys functional organiz-
ation. In order to assess the disconnect, a method was proposed to divide the city into
Functional Clusters. These polygons represent more meaningful units in terms of
urban activity and better reect the current functional character of the urban
Urban areas with clustered economic activities belonging to dierent administrative
entities is one of the key issues currently being faced by many cities. For instance,
todays Sternschanze, an entertainment and nightlife district in the city of Hamburg
was, until 2008, part of three dierent districts (Hamburg-Mitte, Eimsbüttel, and
Altona). Indeed, this area, which has a clustered spatial organization of highly specialized
economic activities, has been recognized as an entity due to the administrative challenges
and lack of consensus between the policies adopted by the three districts. One of the main
challenges of this process has been to delineate the districts limits (Bürgerschaft der Freie
und Hansestadt Hamburg, 2006), which could have beneted from the method proposed
in this study.
In terms of the source of information selected, the geo-located ne grain listing of
urban activities from Google Places has proven to be rather useful for the recognition
of dierent clusters of urban economic activities, services, and facilities. Moreover,
since this crowd-sourced information is constantly updated, changes over time can be
monitored. These updates are valuable for keeping track of possible changes in urban
activity and are useful for informing the implementation of policies and the design of
urban intervention strategies based on the functional organization of the city.
However, using Google Places datasets is not without its challenges. A thorough pre-pro-
cessing of data is required before to using the dataset since this social network was
designed for an entirely dierent purpose. For instance, duplication of places found in
the raw dataset is frequent due to the ease with which users can freely create and/or
edit place registrations. Another challenge of using Google Places datasets is that data
are classied into too many similar categories that complicate detailed analysis. Re-cate-
gorizing data into the APA functional categories has helped to overcome this challenge
and has allowed a dierent granularity level in the analysis.
Indeed, the Functional Clusters derived from the present distribution of economic
activities in Alicante do not correspond to the citys current administrative spatial deli-
mitation. In this case study, although economic activities and services on both sides of an
axis give its urban environment a unique character, currently they belong to two dierent
Administrative Neighborhoods. However, with the proposed Functional Clusters, econ-
omically active axis segments take the place of main centers of activity. Hence, the
boundaries of existing administrative divisions that may have been understood as
edgesbehave more like seams(Jacobs, 1961), and become the heart of the new Func-
tional Clusters, thereby creating a polynuclear structure of neighborhoods with greater
cross-movement of people.
Furthermore, the Functional Clusters obtained presented a more homogeneous distri-
bution and degree of specialization in terms of the quantity and diversity of urban activi-
ties, when compared to the Administrative Neighborhoods (See Figure 5). More
specically, in terms of quantity of urban activities, the two ANs with the most and
least registered places have 3,047 and 30 places, respectively, whereas the FCs showed
a proportionally much smaller dierence, with 1,244 and 114 places, respectively. This
also means that, in terms of diversity, the cluster with the least place typesFC 32
has almost four times more diversity than the neighborhood with the least place types
AN 35. Still more evidence is the fact that there are only 9 FCs (as opposed to the
17 ANs) with fewer than 20 dierent place types.
This was not predictable since the spatial clustering was performed exclusively by
taking into account the proximity between datapoints and not the types of urban activi-
ties. A possible explanation of this diversity in the case study responds to the functional
characteristics of the case study selected: a predominantly mixed-use Mediterranean city.
In any case, this study concurs with previous research and provides additional evi-
dence on the extent to which a city subdivision based on its current functional distri-
bution leads to a better understanding of the type of urban life that is being created
(Cranshaw et al., 2012; Crooks et al., 2015; Kropf, 1996; Noulas et al., 2013; Zhang
et al., 2013). Therefore, dening Functional Clusters and acknowledging the quantity,
distribution, and diversity of economic activity, services, and facilities within them,
can be of great value in informing planning processes and shaping the direction of
urban regeneration policies (Cranshaw et al., 2012; Noulas et al., 2013; Zhang et al.,
2013). For instance, knowing the degree of functional specialization is useful when
designing land-use strategies that guarantee the availability and diversity of urban activi-
ties in a neighborhood, thus encouraging residents to stay and support the livability of its
urban spaces.
The main contribution of this paper to the state of the art is threefold:
the use of Google Places data to dene up-to-date functional divisions of the city
the pre-processing and re-categorization of Google Places listings of economic activity
into benchmark categories that allow an assessment at dierent degrees of granularity
the capacity to identify a degree of specialization and complexity of urban activity (in
terms of quantity and diversity) for each Functional Cluster.
The novelty of the proposed methodological framework lies in two aspects. First, it uses a
social network that specically includes information about places and urban activities
on offer.Therefore, a detailed listing of the available economic activities, services, and
facilities in a given urban area can be obtained. Previous studies whose focus was to delin-
eate urban functional areas using social media data have used more socially driven social
networks that do not deal with the available urban activities in a place, but rather the
demand for activities, that is, where people add a register on the particular social
network (i.e., visit, tweet or check-in). For instance, the work of Cranshaw et al.
(2012) that uses data retrieved from Foursquare and Twitter; and that of Chen et al.
(2017:49) which collects datasets from Tencent, one of the largest online social media
platforms in China.
Second, the grouping of the original Google Places place types into the hierarchical APA
categorization provides a degree of granularity that allows the assessment of the functional
dimension at dierent scales. A more general Level 1 categorization could be used for
regional scale analysis; the Level 2 is, as demonstrated in this study, useful for city scale
assessment; and the Level 3 or its equivalent, Google Places place types, can be considered
for conducting street, and even urban block-scale analysis. The rst and third cases have
not been fully explored, and as such are potential future directions for research.
To sum up, this paper broadens the debate on how to understand the neighborhood
unit, based on todays urban functional reality. The study suggests that existing admin-
istrative boundaries are becoming increasingly obsolete and the use of social network
data cannot be neglected in the present debate on neighborhood units.
1. Script understood as a sequence of instructions carried out by another program.
Notes on Contributors
Pablo Martí is a researcher and professor in the Urban Design and Regional Planning Unit, Uni-
versity of Alicante.
Leticia Serrano-Estrada is a researcher and lecturer in the Urban Design and Regional Planning
Unit, University of Alicante.
Almudena Nolasco-Cirugeda is a researcher and lecturer in the Urban Design and Regional Plan-
ning Unit, University of Alicante.
Jesús López Baeza is a researcher in the CityScienceLab, HafenCity University Hamburg.
This work was supported by the Conselleria de Educación, Investigación, Cultura y Deporte, Gen-
eralitat Valenciana (Spain). Project: Valencian Community cities analyzed through Location-
Based Social Networks and Web Services Data. Ref. no. AICO/2017/018.
This research was funded by Conselleria de Innovación, Universidades, Ciencia y Sociedad Digital,
Generalitat Valenciana, Spain -(GV/2021/177).
Pablo Martí
Leticia Serrano-Estrada
Almudena Nolasco-Cirugeda
Jesús López-Baeza
S. Abdullahi, B. Pradhan, and H. Mojaddadi, City Compactness: Assessing the Inuence of the
Growth of Residential Land Use,Journal of Urban Technology 25: 1 (2018)2146.
American Planning Association, Land Based Classication Standards (LBCS) (2018a)<https://> Accessed January 18, 2018.
American Planning Association, LBCS Function Dimension with Descriptions (2018b)<https:// > Accessed January 18, 2018.
D. Arribas-Bel and E. Tranos, Big Urban Data: Challenges and Opportunities for Geographical
Analysis,Geographical Analysis 50: 2 (2018) 123124.
Ayuntamiento de Alicante, Guía Urbana de Alicante(2018)<>
Accessed April 12, 2018.
J. Béjar, S. Álvarez, D. García, I. Gómez, L. Oliva, and A. Tejeda, Discovery of Spatio-temporal
Patterns from Location-based Social Networks,Journal of Experimental & Theoretical
Articial Intelligence 28: 1-2 (2016) 313329.
S. Brown, Retail Location: The Post Hierarchical Challenge,The International Review of Retail,
Distribution and Consumer Research 1: 3 (1991) 367381.
S. Brown, Micro-Scale Retail Location: Cinderella or Ugly Sister?International Journal of Retail
& Distribution Management 21: 7 (1993)1019.
Bürgerschaft der Freie und Hansestadt Hamburg, Mitteilung des Senats an die Bürgerschaft,
Drucksache 18/5011 (Hamburg: Mitteilung des Senats, 19 September 2006) <
de/contentblob/2240328/c9059f48298bd8c7575371ddd40de15f/data/drucksache-18-5011.pdf >
Accessed June 2, 2020.
M. Bustos Hernándes, Análisis de la Complejidad Urbana en la Ciudad Turística : El Caso de La
Pineda (Vila-seca, Tarragona),paper presented at the International Conference on Regional
Science: Innovation and Geographical Spillovers: New Approaches and Evidence, XLI
Reunión de Estudios Regionales - AECR (Reus, November 18-20, 2015)
I. Cenamor, T. de la Rosa, S. Núñez, and D. Borrajo, Planning for Tourism Routes Using Social
Networks,Expert Systems with Applications 69 (2017)19.
R. J. Chaskin, P. Brown, S. Venkatesh, and A. Vidal, Building Community Capacity, (New York:
Aldine de Gruyter, 2001).
Urban Functional Areas with Building-Level Social Media Data: A Dynamic Time
Warping (DTW) Distance Based K-Medoids Method,Landscape and Urban Planning
160 (2017)4860.
J. Cranshaw, R. Schwartz, J. I. Hong, and N. Sadeh, The Livehoods Project: Utilizing Social Media
to Understand the Dynamics if a City,in 6th International AAAI Conference on Webblogs and
Social Media (Dublin, Ireland: 2012).
A. Crooks, D. Pfoser, A. Jenkins, A. Croitoru, A. Stefanidis, D. Smith, S. Karagiorgou, A. Efentakis,
and G. Lamprianidis, Crowdsourcing Urban Form and Function,International Journal of
Geographical Information Science 29: 5 (2015) 720741.
A. T. Crooks, A. Croitoru, A. Jenkins, R. Mahabir, P. Agouris, and A. Stefanidis, User-Generated
Big Data and Urban Morphology,Built Environment 42: 3 (2016) 396414.
X. Deng and S. Newsam, Quantitative Comparison of Open-Source Data for Fine-Grain Mapping
of Land Use,Proceedings of the 3rd ACM SIGSPATIAL Workshop on Smart Cities and Urban
AnalyticsUrbanGIS17 (2017)18.
Density Design Research Lab, RAW Graphs (2018) <> Accessed March 16,
D. Devine, Neighborhood Unit Concept (1975). <
555329> Accessed June 2, 2020
I. G. Ellen and M. A. Turner, Does Neighborhood Matter? Assessing Recent Evidence,Housing
Policy Debate 8:4 (1997) 833866.
S. Fortunato, Community Detection in Graphs,Physics Reports 486: 3-5 (2010)75174.
F. F. Furstenberg, How Families Manage Risk and Opportunity in Dangerous Neighborhoods,in
W. J. Wilson, ed., Sociology and the Public Agenda, American Sociological Association
Presidential Series (London: Sage, 1993).
J. C. García-Palomares, J. Gutierrez, and C. Mínguez, Identication of Tourist Hot Spots Based on
Social Networks : A Comparative Analysis of European Metropolises Using Photo-Sharing
Services and GIS,Applied Geography 63 (2015) 408417.
J. C. García-Palomares, M. H. Salas-Olmedo, B. Moya-Gómez, A. Condeço-Melhorado, and J.
Gutiérrez, City Dynamics Through Twitter: Relationships Between Land Use and
Spatiotemporal Demographics,Cities 78(Part B) (February, 2018) 310319.
Google Developers, Supported Place Types (2018)<
supported_types> Accessed May 10, 2017.
S. Graham and S. Marvin, Splintering Urbanism: Networked Infrastructures, Technological
Mobilities, and the Urban Condition (London-New York: Routledge, 2002).
K. Gu, Urban Morphological Regions and Urban Landscape Management: The Case of Central
Auckland, New Zealand,Urban Design International 15: 3 (2010) 148164.
C. Guy, The Location of Shops in the Reading Area (Geographic) (Reading: University of Reading.
Department of Geography, 1976).
H. W. Hallman, Neighborhoods: Their Place in Urban Life, Sage Library of Social Research (Beverly
Hills, CA: Sage, 1984).
N. Hossain, A Syntactic Approach to the Analysis of Spatial Patterns in Spontaneous Retail
Development in Dhaka,Space Syntax Second International SymposiumProceedings I(1999).
J. Jacobs, The Death and Life of Great American Cities (New York: Vintage Books, 1961).
M. Jenks and N. Dempsey, Dening the Neighbourhood: Challenges for Empirical Research,
Town Planning Review 78: 2 (2007) 153177.
M. Kärrholm, Retailing Space. Architecture, Retail and the Territorialisation of Public Space
(Surrey: Ashgate, 2012).
S. S. Khan and A. Ahmad, Cluster Center Initialization Algorithm for K-Means Clustering,
Pattern Recognition Letters 25 (2004) 12931302.
K. Kropf, Urban Tissue and the Character of Towns,Urban Design International 1: 3 (1996) 247
Y. Lee and M. McCracken, Spatial Adjustment of Retail Activity: A Spatial Analysis of
Supermarkets in Metropolitan Denver, 19601980,Journal of Regional Analysis and Policy
12:2 (1982)6276.
J. López Baeza, D. Cerrone, and K. Männigo, Comparing Two Methods for Urban Complexity
Calculation Using Shannon-Wiener Index,in WIT Transactions on Ecology and
Environment (Ashurst: Ashurst Lodge, 2017).
K. Lynch, The Image of the City (Cambridge MA: MIT Press, 1960).
E. Mankalpa, C. Marmolejo, and M. S. Far, An Overview of Restoration and Regeneration
Possibilities of Spanish Mediterranean Coastal Zones; Case of Second Home Tourism as
Retirement Plans,presented at the European Regional Science Association ERSA 55th
Congress (Lisbon, August 25-28, 2015)
P. Martí, L. Serrano-Estrada, and A. Nolasco-Cirugeda, Using Locative Social Media and Urban
Cartographies to Identify and Locate Successful Urban Plazas,Cities 64 (2017)6678.
P. Martí, L. Serrano-Estrada, and A. Nolasco-Cirugeda, Social Media Data: Challenges,
Opportunities and Limitations in Urban Studies,Computers, Environment and Urban
Systems 74 (March 2019) 161174.
D. Martin, Automatic Neighbourhood Identication from Population Surfaces,Computers,
Environment and Urban Systems 22: 2 (1998) 107120.
V. Mehta and B. Mahato, Measuring the Robustness of Neighbourhood Business Districts,
Journal of Urban Design 24: 1 (2018)99118.
A. Mubi Brighenti and M. Kärrholm, Atmospheres of Retail and the Asceticism of Civilized
Consumption,Geographica Helvetica 73: 3 (2018) 203213.
A. Nejat, Perceived Neighborhood Boundaries: A Missing Link in Modeling Post-Disaster
Housing Recovery,International Journal of Disaster Risk Reduction 28 (June 2018) 225236.
E. Nieto, Commons-based Urbanism: Can Alicante be a Case Study?in J. Almazán, ed., Post-
Souvenir City. Mediterranean Urban Intensity and New Tourism Practices in Alicante (Tokyo:
IKI -International Keio Institute and Flick Studio, 2015).
A. Noulas, C. Mascolo, and E. Frias-Martinez, Exploiting Foursquare and Cellular Data to
Infer User Activity in Urban Environments,paper presented at 2013 IEEE 14th
International Conference on Mobile Data Management 1 (2013). <
N. A. Omar, Quality of StreetsQuality of Urban Life: A Case Study of Tanta City, Egypt Engy
H. Saeed,Journal of Urban Research 31 (January 2019)79102.
C. Perry, The Neighborhood Unit : Neighborhood and Community Planning of the Regional Survey
of New York and Its Environs (New York: Committee on Regional Plan of New York and Its
Environs, 1929).
W. M. Rohe, From Local to Global. One Hundred Years of Neighborhood Planning,Journal of
the American Planning Association 75: 2 (2009) 209230.
R. Rösler and T. Liebig, Using Data from Location Based Social Networks for Urban Activity
Clustering,in D. Vandenbroucke, B. Bucher, and J. Crompvoets, eds., Geographic
Information Science at the Heart of Europe: Lecture Notes in Geoinformation and Cartography
(Switzerland: Springer, 2013).
M. Saraiva, The Morphological Sense of Commerce. Symbioses between Commercial Activity and the
Form and Structure of Portuguese Medium Sized Cities, Dissertation, University of Porto (2013)
M. Saraiva and P. Pinho, Spatial Modelling of Commercial Spaces in Medium-Sized Cities Cities,
GeoJournal 82 (2017) 433454.
A. K. Sarma, The Social Logic of Shopping: A Syntactic Approach to the Analysis of Spatial and
Positional Trends of Community Centre Markets in New Delhi,MastersThesis (2006)
H. Serag El Din, A. Shalaby, H. E. Farouh, and S. A. Elariane, Principles of Urban Quality of Life
for a Neighborhood,HBRC Journal 9: 1 (2013)8692.
L. Serrano-Estrada, P. Marti, and A. Nolasco-Cirugeda, Comparing Two Residential Suburban
Areas in the Costa Blanca, Spain,Articulo. Journal of Urban Research 13-Suburbia (2016).
Y. Shen and K. Karimi, Urban Function Connectivity: Characterisation of Functional Urban
Streets with Social Media Check-in Data,Cities 55 (2016)921.
Spanish Statistics Institute. Cifras oOciales de Población Resultantes de la Revisión del Padrón
Municipal a 1 de Enero 2017. Alicante: Población por Municipios y Sexo (2018)<https://>
E. Steiger, B. Resch, and A. Zipf, Exploration of Spatiotemporal and Semantic Clusters of Twitter
Data Using Unsupervised Neural Networks,International Journal of Geographical Information
Science 8816: July (2015)123.
P. Suau-Sanchez and G. Burghouwt, Connectivity Levels and the Competitive Position of Spanish
Airports and Iberias Network Rationalization Strategy, 20012007,Journal of Air Transport
Management 18: 1 (2012)4753.
R. R. Temes Cordóvez, M. R. Simancas Cruz, M. P. Peñarrubia Zaragoza, A. Moya Fuero, and A.
M. García Amaya, Characterization and Spatial Identication of Holiday Tourist Assessments
in the City of Valencia,in J. Rivas Navarro and B. Bravo Rodríguez, eds., 6th Sustainable
Development Symposium: Book of Abstracts (Granada: Godei, 2016).
The Urban Task Force, Towards an Urban Renaissance. Final Report of the Urban Task Force
(London: Taylor & Francis, 2003).
S. Van Canneyt, S. Schockaert, O. Van Laere, and B. Dhoedt, Detecting Places of Interest Using
Social Media,in 2012 IEEE/WIC/ACM International Conferences on Web Intelligence and
Intelligent Agent Technology (Macau, December 4-7, 2012)<
D. Van Weerdenburg, S. Scheider, B. Adams, B. Spierings, and E. Van Der Zee, Where to Go and
What to Do: Extracting Leisure Activity Potentials from Web Data on Urban Space,
Computers, Environment and Urban Systems 73 (2019) 143156.
A. X. Zhang, A. Noulas, S. Scellato, and C. Mascolo, Hoodsquare: Modeling and Recommending
Neighborhoods in Location-Based Social Networks,2013 International Conference on Social
Computing (Alexandria, September 8-14, 2013)<>
S. Zukin, Landscapes of Power (Los Angeles: University of California Press, 1993).
... Spatial configuration measures such as integration, connectivity and choice although equivalent to closeness, degree and betweenness (Hillier and Iida, 2005;Omer and Goldblatt, 2016), have hardly been used in location modelling. Nevertheless, they have proven significant at predicting economic activity, and specifically retail location (Hillier, 1996;Lin et al., 2018;Martí et al., 2021;Mora, 2003;Ortiz-Chao and Hillier, 2007). The poor assimilation of spatial configuration measures by location modelling can eventually be attributed to the fact that, as measures become more abstract, their association with traffic flow modelling becomes less obvious, and this has been one of the main goals of location modelling (Batty, 2009). ...
Full-text available
Although location theory is now almost two centuries old, the firms' location choice processes are yet to be fully understood. And while accessibility, in some form, has long been used as an explanatory factor, spatial configuration measures (space syntax) have hardly been used in location models, and longitudinal analyses have also been infrequent. Therefore, a panel multinomial logit model is implemented in Lisbon to explore the role of these specific type of accessibility measures in shaping firm location, throughout a 15-year period, using data from 1995, 2002 and 2010. The analysis is focused on retail activity (five retail categories), restaurants and cafes. The results show that firm location can persistently be related to Local Integration and Choice, although the magnitude of the relationship changes depending on the commercial category. These results reinforce the general idea that good urban design can potentially counteract information costs and help commerce locate in new areas of the city while allowing it to continue to succeed in the city's older, more central areas.
... Finally, the enrichment of the data provided by Google Places Enricher can help different studies interested in data of establishments with detailed categories, such as the work by Martí et al. [9]. Promoting the debate on the spatial definition of neighbourhood boundaries, the authors conducted a study using Google Places data. ...
Conference Paper
The growing number of available APIs means that more developers and interested users need to learn to use unfamiliar interfaces, requiring a learning curve that can compromise productivity. Thus, it becomes important to find ways to facilitate their usability. This work presents a tool that facilitates using the Google Places API, simplifying multiple API calls to cover a region of interest. In addition, the proposed tool also provides features for the enrichment of these data, extending the PoI data from that region with categories from other sources. It is hoped that developers and users without much computer knowledge can benefit from Google Places Enricher, helping to ease the development of new sophisticated urban applications and services.
... Theoretically, urban expansion within a gridcell started with an increase in urban impervious surface (UIS), which reduce urban-rural vegetation (URV) [21]. Urban land management also affects URV, including (i) land cover change (such as new lawns, open parks, and green-belts); (ii) vegetation species change and growth in suburban, periurban, and neighboring rural areas; and (iii) vegetation management (such as fertilization and irrigation) [29,30]. In recent years, many cities started to improve and expand urban vegetation and considered urban vegetation as one of major metrics for the society to address the ecological and environmental impacts of urbanization [31]. ...
Full-text available
Urbanization affects vegetation within city administrative boundary and nearby rural areas. Gross primary production (GPP) of vegetation in global urban areas is one of important metrics for assessing the impacts of urbanization on terrestrial ecosystems. To date, very limited data and information on the spatial-temporal dynamics of GPP in the global urban areas are available. In this study, we reported the spatial distribution and temporal dynamics of annual GPP during 2000–2016 from 8,182 gridcells (0.5° by 0.5° latitude and longitude) that have various proportion of urban areas. Approximately 79.3% of these urban gridcells had increasing trends of annual GPP during 2000-2016. As urban area proportion (%) within individual urban gridcells increased, the means of annual GPP trends also increased. Our results suggested that for those urban gridcells, the negative effect of urban expansion (often measured by impervious surfaces) on GPP was to large degree compensated by increased vegetation within the gridcells, mostly driven by urban management and local climate and environment. Our findings on the continued increases of annual GPP in most of urban gridcells shed new insight on the importance of urban areas on terrestrial carbon cycle and the potential of urban management and local climate and environment on improving vegetation in urban areas.
... Recent research has taken advantage of new big data sources for the intraurban analysis of consumption patterns, for example, using Google points of interest (Martí Ciriquián et al., 2021) and user-generated data from social networks (Carpio-Pinedo & Gutiérrez, 2020;Salas-Olmedo et al., 2018). However, bank card data has a number of advantages worth of further investigation. ...
Full-text available
The spatial distribution of commercial activities is vital to support healthy lifestyles and to achieve livable public spaces and environmental, social and economic sustainability in our cities. However, commercial activities require a constant flow of expenditure for their own viability. As a result, understanding the spatial and temporal distribution of expenditure is fundamental, although the lack of detailed, complete data sources has impeded this task until now. Bank card data paves the way for a new urban geography of expenditure, thanks to its fine spatial and temporal granularity along with the uniform coverage of all commercial sectors. In this paper, we analyze temporal, spatial, and spatiotemporal distributions of expenditure at the intraurban scale of the city of Madrid (Spain), combining spatial statistical tools (Getis-Ord General for global autocorrelation and Getis-Ord Gi* hot spot analysis for local autocorrelation) with k-means cluster analysis and spatiotemporal tools (Time Series Clustering analysis and Temporal Hot Spot Analysis). Our analysis confirms the strong center-periphery gradient described in previous literature, but with a CBD integrated by distinct specialized areas. The paper demonstrates that bank card data has a great potential to support a new geography of expenditure that could strengthen decision-making in planning and retailing.
... The consequent pedestrian flow produced by the amenities and services follows certain rules, in our two use cases, in which higher density and diversity of amenities tend to produce a higher density of pedestrians walking predominantly on a local scale, potentially producing a higher sense of locality in line with Martí et al. [51] describing urban clusters of social activity following the aggregation of economic activity. The results presented in this research not only validate their approach but also quantify it with measurable indicators. ...
Full-text available
Pedestrian activity is a cornerstone for urban sustainability, with key implications for the environment, public health, social cohesion, and the local economy. Therefore, city planners, urban designers, and decision-makers require tools to predict pedestrian mobility and assess the walkability of existing or planned urban environments. For this purpose, diverse approaches have been used to analyze different inputs such as the street network configuration, density, land use mix, and the location of certain amenities. This paper focuses on the location of urban amenities as key elements for pedestrian flow prediction, and, therefore, for the success of public spaces in terms of the social life of city neighborhoods. Using agent-based modeling (ABM) and land use floor space data, this study builds a pedestrian flow model, which is applied to both existing and planned areas in the inner city of Hamburg, Germany. The pedestrian flows predicted in the planned area inform the ongoing design and planning process. The flows simulated in the existing area are compared against real-world pedestrian activity data for external validation to report the model accuracy. The results show that pedestrian flow intensity correlates to the density and diversity of amenities, among other KPIs. These correlations validate our approach and also quantify it with measurable indicators.
Full-text available
La situación de emergencia sanitaria COVID-19 ha impactado múltiples dimensiones de la compleja estructura física, social, funcional y económica de las ciudades. Esta investigación lleva a cabo un diagnóstico comparado sobre algunos cambios y transformaciones que se han producido en el entorno urbano a causa de la crisis y que se reflejan en los datos de redes sociales geolocalizadas. Para ello, los datos de Google Places y Twitter se adoptan como principal fuente de información. Se propone un método mixto, cualitativo y cuantitativo, para analizar el aumento y la pérdida de actividad económica (Google Places) y presencia humana (Twitter) en dos periodos, pre y pos-pandemia. Como caso de estudio, se analizan dos ámbitos con condiciones socioeconómicas muy diferenciadas en tres ciudades localizadas en países que han adoptado distintas medidas de contención de la pandemia –Valencia en España; Ciudad de México en México y Gotemburgo en Suecia–. El diagnóstico informado por estas redes sociales es de gran utilidad para formular estrategias útiles tanto para identificar los cambios que se han venido produciendo como para afrontar futuros escenarios disruptivos.
Full-text available
La revolución de los datos masivos —Big Data— ha irrumpido con nuevas fuentes de datos y novedosas metodologías en el estudio de la ciudad. Entre estos datos sobresalen aquellos que están geolocalizados y, por tanto, ofrecen información precisa sobre los lugares de la ciudad en los que se generan o comparten; en este sentido, destacan los datos geolocalizados de las redes sociales. Esta información permite evaluar aspectos relacionados con la sostenibilidad de las ciudades y sus comunidades de acuerdo con los Objetivos de Desarrollo Sostenible—. A través de tres escalas de estudio —la ciudad, el barrio y el espacio público local— se abordan aquellas cuestiones que informan sobre: centralidad urbana, equilibrio de los distintos sistemas urbanos, densidad y distribución de actividades económicas y urbanas, presencia y preferencias ciudadanas y vitalidad en espacios públicos urbanos.
Full-text available
Analysing the city through data retrieved from Location Based Social Networks (LBSNs) has received considerable attention as a promising method for applied research. However, the use of these data is not without its challenges and has given rise to a stream of polemical arguments over the validity of this source of information. This paper addresses the challenges and opportunities as well as some of the limitations and biases associated with the collection and use of LBSN data from Foursquare, Twitter, Google Places, Instagram and Airbnb in the context of urban phenomena research. The most recent research that uses LBSN data to understand city dynamics is presented. A method is proposed for LBSN data retrieval, selection, classification and analysis. In addition, key thematic research lines are identified given the data variables offered by these LBSNs. A com- prehensive and descriptive framework for the study of urban phenomena through LBSN data is the main con- tribution of this study.
Full-text available
Web data is the most prominent source of information for deciding where to go and what to do. Exploiting this source for geographic analysis, however, does not come without difficulties. First, in recent years, the amount and diversity of available Web information about urban space have exploded, and it is therefore increasingly difficult to overview and exploit. Second, the bulk of information is in an unstructured form which is difficult to process and interpret by computers. Third, semi-structured sources, such as Web rankings, geolocated tags, check-ins, or mobile sensor data, do not fully reflect the more subtle qualities of a place, including the particular functions that make it attractive. In this article, we explore a method to capture leisure activity potentials from Web data on urban space using semantic topic models. We test three supervised multi-label machine learning strategies exploiting geolocated webtexts and place tags to estimate whether a given type of leisure activity is afforded or not. We train and validate these models on a manually curated dataset labeled with leisure ontology classes for the city of Zwolle, and discuss their potential for urban leisure and tourism research and related city policies and planning. We found that multi-label affordance estimation is not straightforward but can be made to work using both official webtexts and user-generated content on a medium semantic level. This opens up new opportunities for data-driven approaches to urban leisure and tourism studies.
Full-text available
p>During recent decades, consumption-oriented spaces of comfort and hospitality have proliferated - including, for instance, lounge shopping malls, food court plazas, spas, entertainment retail, visitor centres, and the development of ever larger pedestrian precincts. In this article we explore shopping malls as capitalist domes in Sloterdijk's sense. We observe atmospheric production, atmospheric management and atmospheric culture (which we propose to call atmoculture ) inside such domes. Processes of retailization and mallification - whereby shopping malls and retail spaces absorb increasing economic and societal energies - can be regarded as correlative to the rise of an atmoculture of civilized consumption. Such atmoculture is visible for instance in stress-avoidance strategies and the production of a pleasurable experience in consumption-oriented public zones. The design of contemporary retail spaces seems to pivot around specific atmospheric strategies developed to promote and sustain civilized consumption. In this piece, we describe four different strategies of atmospheric production, identifying their possible shortcomings and failings. Finally, we advance the hypothesis that the atmospheric production of retail can also be analyzed with reference to Sloterdijk's theorization of asceticism as self-disciplination. .
Conference Paper
Full-text available
This paper performs a quantitative comparison of open-source data available on the Internet for the fine-grain mapping of land use. Three points of interest (POI) data sources--Google Places, Bing Maps, and the Yellow Pages--and one volunteered geographic information data source--Open Street Map (OSM)--are compared with each other at the parcel level for San Francisco with respect to a proposed fine-grain land-use taxonomy. The sources are also compared to coarse-grain authoritative data which we consider to be the ground truth. Results show limited agreement among the data sources as well as limited accuracy with respect to the authoritative data even at coarse class granularity. We conclude that POI and OSM data do not appear to be sufficient alone for fine-grain land-use mapping.
A key component of vibrant urban neighbourhoods is the neighbourhood business district (NBD) – the place for amenities, shopping, social engagement and recreation. In order to provide these advantages and be competitive with big box and online retailers, the NBD must be easily accessible and provide a basic range of goods and services. Yet few tools exist that measure both the access and completeness of amenities on NBDs. This paper presents a systematic analysis and provides cities with a tool to assess NBDs and create policies and programmes to support more complete, robust and useful neighbourhood business districts.
Literature in the domain of disasters has highlighted the importance of reestablishment of housing to restore a community due to its ripple effect on the overall timing of recovery. In fact, previous research studies have revealed the significance of members of social support networks, such as neighbors, church members, and others who share local communal spaces, on the making of recovery decisions. The objective of this research is to explore the determinants that drive households’ perceived neighborhood boundaries, which will in turn facilitate future research on exploring the importance of community assets which would lead to establishment of these social support networks within perceived neighborhoods. To accomplish this, an online survey was designed and conducted in the states of New York and Louisiana, which bore the brunt of the nation's two costliest hurricanes, Sandy and Katrina. The survey provided an online graphical user interface with which participants could manually draw a polygon around the area they perceived to be their neighborhood. These manually-selected neighborhoods were then compared to fixed units defined by the Census Bureau to check for similarities and differences. In summary, out of various internal attributes, prior disaster-related personal impact resulted in identifying larger areas as perceived neighborhood boundaries. Similarly, location density was shown to have an increasing effect on perceived neighborhood area. Moreover, married participants tended to have larger perceived neighborhoods when compared to their unmarried counterparts. Finally, the results revealed a direct relationship between number of perceived community assets and size of perceived neighborhood.
In the urban sprawl paradigm, residential land use exhibits a more significant growth than other categories. Consequently, large proportions of the natural environment are converted to residential areas, particularly in tropical countries. Compact urban development is one of the most sustainable urban forms with environmental perspectives, such as rural development containment and natural environment preservation. However, no proper investigation of the relationship and influence of residential growth and city compactness is available. This study evaluated and forecasted the residential development of Kajang City in Malaysia based on compact development. First, the relationship between residential land use change and city compactness was evaluated. Second, residential growth was projected by utilizing the land transformation model (LTM) and the statistical-based weight of evidence (WoE) using various spatial parameters. Both models were evaluated with respect to observed land use and compactness maps. Results indicated that most of the newly developed residential areas were in zones where the degrees of compactness increase during certain periods. In addition, LTM performed better and provided a more accurate modeling of residential growth than the WoE. However, WoE provided clearer and more informative results than LTM in terms of functional relationships between dependent and independent variables related to city compactness.