Content uploaded by Jesús López Baeza
Author content
All content in this area was uploaded by Jesús López Baeza on Oct 11, 2017
Content may be subject to copyright.
CAI S A CI
CACAI SI SAI I
SS AA,, AIA C, ISA I
niversity of Alicante, Spain
ampere niversity of echnoloy, inland
SI nit, stonia
ASAC
his study will compare the results of measurin rban Compleity usin the Shannoniener inde
in two different methods. sin a oint dataset retrieved from oursuare AI, we will measure the
deree of urban compleity of every street relatin every amenity to the closest street sement in a
computational way and then applyin the calculation to the sements and applyin the calculation
to every cell of a rid that will be combined with the street networ afterwards. he selected case study
is the city of ondon, and the dataset employed will be retrieved from oursuare. ver , venues
were collected and classified in over cateories. In order to proceed to the analysis, these
cateories will be reduced to based on the classification of activities observed in the public space
from the traditional urban discipline. hen the urban compleity inde of each street sement of ondon
will be measured as a simultaneous calculation of the density and diversity of collected and classified
economic activities.
Keywords: urban complexity, Shannon-Wiener, social media, London.
ICI
he use of social media data when analysin the city is nowadays becomin popular in the
urban plannin discipline . IS technoloies allow us not only to visualie but also operate
with data retrieved from social networs and online servers. urthermore, due to the
populariation of social media amon smartphone users, data samples are reater in surface
and in number of points than any other data sources so far , and they have a hiher
spatial and temporal resolution .
In this paper, we aim to study the intanible relations between city form and the
distribution of activities hosted by amenities. e speculate that compact urban forms foster
a hiher variety of amenities capable to host a mi of different activities, which are beneficial
to life in the city. erceivably dense urban areas with small narrow streets such as uropean
old towns enerally cluster a hih amount of venues with a hih level of compleity.
Scholars have attempted to measure the urban compleity inde throuh the application
of standardied indees. Specifically, this article uses the Shannoniener inde tain
points relative to venues reistered in oursuare for the calculation. his data source has
been utilied throuh the socalled Grid calculation method, capable of determinin an inde
of compleity for a iven area divided in eual cells . otwithstandin, we aim to measure
the compleity within the street. hus, in this paper we will obtain a value of urban
compleity attached to every street sement of the street networ of ondon city refinin the
calculation for a iven surface two dimensional into a unit of lenth one dimension. o
do so, we will test two different methods Closest oint C and rid method.
ORCID: http://orcid.org/0000-0002-4092-1782
ORCID: http://orcid.org/0000-0002-4853-9887
ORCID: http://orcid.org/0000-0002-6592-9588
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
Sustainable Development and Planning IX 369
doi:10.2495/SDP170321
e epect to obtain similar results. owever, when measurin with Closest oint C
calculation these venues would be divided in smaller roups relatin every one of them to
the closest street and the compleity value would be lower. n the contrary, sprawl fabric
with a low density of amenities miht obtain a low value of compleity when measured
throuh rid calculation, but a hiher value when measured throuh C calculation, since
points would be lined to the closest street without tain a maimum distance into account.
. lacebased social media
e can define placebased social media as those social networs that record and store the
contents shared on their platforms and the eoraphic location they were shared from. his
data can often be accessed by third parties from AI services. Since our research focuses on
the metadata reardin the contents location rather than the personal information of users,
our methodoloy is about which place people are postin from rather than who is postin.
he use of placebased applications in spatial plannin research is enerally lined to the
study of landscape or in particular, to observe a particular area in order to uantify the impact
of certain factors to describe intanible urban phenomena.
lacebased social media such as oole laces has been used by urban studies scholars
as a tool to measure urban compleity , , due to its ability to represent the totality
of businesses reistered in oole with eoraphic coordinates.
In this way, placebased social media offer a virtual representation of the heteroeneity of
the economic activities of a certain area, constantly updated and revised by the users
themselves. In addition to location, they offer other information such as name, address, and
business cateory , allowin the classification of activities accordin to type. Althouh
this data can be easily accessed and it is storin the eact eoraphic coordinates of
businesses, the reliability on the content itself it may be hard to verify.
. oursuare
In this study, we are oin to focus on oursuare, a social media platform that allows users
to add enues worth visitin and epressin preferences. he app also allows checins
for users who intend to state their presence in a specific venue. riinally oursuare was
desined to be used as a ame, where users with most checins in one venue could earn
bades and popularity . hile the oursuare mobile app still allows users to checin
and review venues, the ludic component has been removed from the service. In the past years,
venues were mostly trendy places such as bars, restaurants and clubs, but only recently
oursuares coverae because broader and more detailed. oday the list of venues is no
loner limited to few urban amenities but to the overall set of locations, places and spaces
present in one city includin open areas, infrastructures and buildins hostin real estate or
small businesses.
oursuare is understood by omninos et al. as a collective nowlede tool throuh
which users use information from third parties when main decisions about areas of interest
in the urban contet such as how to spend their free time ., what venues its worth
visitin ., and when to o to visit the selected venues . here are several
studies that discuss oursuare users location choice and their relation to the amelie
interface capable of influencin individuals decision main in the city . In
addition to the proress that users can mae by checinin venues and leavin reviews,
users behaviour is an aspect of the individual spatial self, which can be interpreted as a
representation of lifestyle in order to create or maintain social relationships .
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
370 Sustainable Development and Planning IX
he base of this platform is a series of public or private places listed by users themselves,
in which anyone can reister a visit by checinin . hese places are classified throuh
cateories that are predetermined by the application. hus, data obtained from the searches
throuh this app are liely to be filtered .
he main implementation of oursuare in the field of urbanism is framed in the ability
to list places and reister them with a diital identification and coordinates, and addin
metadata to every reistered place. Specifically, when measurin user preferences for both
oursuare application and the physical space of the city, we will place particular emphasis
on data about the number of visitors and checins. In other words, oursuare, with its
particular amelie interface allows the user to reveal where heshe has been and how many
times. In this way, the information obtained from this app allows to understand human
mobility in its spatial, temporal, social, and content aspects , p. .
here are several articles that relate the use of the oursuare to the observation of the
diversity of economic activities in a iven area listin reistered places and observin
distribution patterns.
he dataset retrieved from the oursuare AI corresponds to every point obtained in the
selected uery area would correspond to a physical place facility, venue, amenity in the
urban environment reistered by the app. In this way, for each place we obtain the followin
data
. Spatial coordinates atitude and onitude
. usiness ame Identification
. Cateory and Subcateories
. Address, street and number
. Averae ratin
. umber of visits or different users who have visited the site one visit per user
. umber of checins or sum of all visits that all users have made onsite.
CAIS ACS
he eoraphic distribution of amenities is topical to the definition of spatiotemporal
patterns of urban and rural settlements. umerous studies have revealed the spatial relations
that overn the location choice of these establishments althouh it is primal to map and
understand these relations in their sociocultural contet , . In particular we are oin
to focus on clusterin and dispersal phenomena to learn how the spatial relations between
socioeconomic establishments are typical to the urban and reional contets of one city.
ehl et al. maes a classification of observed activities performed by people within
the urban space. hese activities were classified by nature in ptional, ecessary and
Social bein ptional those performed when one has free time ym, museum, spa, etc.,
ecessary those interated in the daily routine school, office, maret, etc. and Social
would include those activities that involve interaction with others bar, caf, club, etc. .
In this contet, we may understand Social activities as a part of those performed when one
has free time, so we included them into the cateory ptional. hus, we firstly divide all
activities in these two maor roups, ecessary and ptional.
In order to be able to operate with cateories in a more detailed way and define uses, it is
needed to define activity types more accurately, but at the same time to let these new
cateories be fleible enouh to include all ind of oursuare venues in them. o do so we
use the SI nit rban activity heel .
he novelty of this approach is the classification of amenities not by type or function but
accordin to the spectrum of human activities they can host. In this way, we are oin to
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
Sustainable Development and Planning IX 371
perform an indirect observation of activity patterns that can be scaled from local to reional
scale somethin that is traditionally performed with lon terms and microscale only
surveys a field observation.
ollowin this discourse, we created cateories accordin to the SI nit rban activity
heel i. .
his classification method was created from empirical observations of people performin
activities in public and private spaces to study the difference between what the city offers
Supply throuh its amenities and what people do in the city emand. his method is
desined to cateorie oursuare venues to map and measure the Supply and other datasets
such as Instaram and witter to measure human activity patterns in the city emand. In
this paper, we reclassified oursuare data only to measure the compleity of activity
patterns offered by the city and study if their distribution has a clear relation to the perceived
form of the city.
In this paper ver , venues were collected and classified in over cateories. In
order to proceed to the analysis, these cateories were reduced to based on the SPIN
Unit Urban Activity Wheel i. .
A CI
he concept of Compleity in urban studies starts to be related to the economic activities in
urban spaces reistered in online applications and web services in the mids .
owever, a standardied methodoloy to measure it has not been developed yet. o do so,
first of all, it will be necessary to define this concept and identify methods throuh which it
can be observed and measured.
ther previous authors such as ui Snche , p. have defined the city as a complex
self-regulating system in which all the elements that shape it in their multiple scales are
related to each other in a comple networ, in which the minimum modification in one of
its elements will be a readustment on a lobal scale. lanco and Subitrats associate
this heteroeneous compleity with a real and potential conflict an effect of increasin
sie and conseuent compleity of cities, main them unable to be manaed , or
becomin a challene per se for policy maers .
iure SI nit rban activity heel describin activities in urban space .
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
372 Sustainable Development and Planning IX
iure Cateories in oursuare reclassified into Urban Activity Wheel of activities in
urban space. (Self-elaborated.)
rom a pramatic point of view, ueda refers to compleity as the measure of the
degree of organization of the urban system in terms of the diversity of the mix of uses and
services , p. . Compleity defined as the set of relations between all elements that
constitute the city, the comple selfreulatin system shows the ecosystemic nature of the
same. or this reason, ueda states that this deree of compleity can be obtained
throuh the application of the euation correspondin to the Shannoniener compleity
inde developed to measure the deree of entropy of an ecosystem .
lorida will later relate this urban compleity of cities to economic rowth throuh
what he calls the creative class, which he places in specific spaces characteried by
tolerance, openness, and creativity of business . o this end, he elaborates a series of
indicators capable of measurin particular aspects that directly influence what he calls social
compleity.
ithin the field of urban studies and in relation to social networs and web services,
olascoCirueda and arcaayor use data collected from oole laces to evaluate
the functional urban compleity of economic activities that directly affect urban space
, p. . hey concluded that the functional appeal lined to the tourism eperience and
indicators of sustainability in its case study enidorm, Spain is proportional to the deree
of compleity throuh the application of the Shannoniener inde. herefore, they also
corroborate the possibility of applyin this inde as a truthful indicator of urban compleity.
ustos ernndes also deals with compleity within the field of urbanism, drawin
relationships between it and concepts such as density, diversity or specialiation with
specific cases of urban reeneration. ie olascoCirueda and arcaayor , he
relates the improvement of the space eperience with the increase of activities and relations
resultin from the increase of urban compleity .
herefore, we could define of urban compleity as a simultaneous measure of both the
density and diversity of economic activities in a iven urban space, which is directly related
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
Sustainable Development and Planning IX 373
to spatial eperience and can be measured uantitatively throuh the application of the
Shannoniener diversity inde.
he followin section will eplain the two methods used for calculatin urban compleity
by applyin the Shannoniener inde usin location data retrieved from oursuare. he
first method rid method divides the land into a rid of cells and calculates for each cell
an inde of compleity that will later be transferred into every street sement contained in
the cell. he second method Closestoint method relates every eolocation to the nearest
street sement and calculates the compleity inde for the street sement per unitary lenth
value. he results obtained from both methods will be interpreted and visualied throuh a
IS platform from which maps will be plotted.
he formula employed for calculation is iven below, where rban compleity
inde S number of cateories pi number of amenities with cateory i divided by
number of amenities. ∑
.
. rid method
After every point is recateoried usin the Urban Activity Wheel of cateories as a
reference, points are laid out on a map usin the location data stored as metadata in form of
eoraphic coordinates. Subseuently, the area the city of ondon in our case is divided
into a rid where each cell has dimensions of meters. he formula used does not
refer directly to any unit of area or lonitude. his rid is the basis for comparin cells with
others with one same dimension. In this paper, it has been taen the value of meters
as unitary tain as a precedent the wor of olascoCirueda and arcia ayor . ence,
oursuare points are bounded inside every rid cell boundary with their cateory values. If
no points were found in a cell, it would be removed from further calculation and assined a
null value. As noted previously, the formula taes into account the number of cateories and
the number of items for each cateory inside every cell. After runnin the Shannoniener
inde calculation, the values obtained are transferred from every cell to every street sement
comprised within that cell. he results obtained will mirror the value of urban compleity in
every street sement per unitary measure of area.
. Closest oint C method
he closest point has accessibility information involved, althouh not precise all the time.
he closest street to any venue is not always the one from which one can access. owever,
it is true in most of the cases. herefore, urban compleity inde calculated throuh C
method in an ideal scenario, would measure the level of compleity of one street sement
reardin all the venues to which one can access from that sement.
Similarly, as described above, we firstly need to reclassify the cateories and plot de
dataset in the map throuh the eoraphical coordinates attached to every oursuare point
as metadata. he street networ with which we will operate should be larer than the area
covered by oursuare points to avoid edge effect in which points located on the ede or near
the edes are findin their closest street only from inside of the clippin boundary whereas
the actual closest street could be located outside the operation area. irstly, we calculate the
closest street sement for every oursuare point in the dataset.
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
374 Sustainable Development and Planning IX
he results of this calculation are displayed for every street sement containin the
information about the amount of oursuare points related to it. In similarity with the
previous method described, all streets unrelated to any point are removed from the calculation
and assined a null compleity value. ue to the variety of lenth between sements, the
compleity value should be divided by the lenth of the sement in order to obtain a unitary
value. hus, the Shannoniener formula is applied for every street sement and the results
obtained will represent the value of compleity per unitary measure of lenth.
SS
After the application of both calculation methods, the results obtained have been represented
in two comparative maps. he case study represented is the city of ondon, .
he center of the city confined between eents par, yde ar and ictoria par,
toether with the south ban of the hames presents a hiher urban compleity inde than
the rest of the study area in both cases. In addition, a series of commercial aes arraned
radially present the hih presence of establishments of commercial activity. his ais is
common to both results and they conform a networ in which the nodes i.e. lephant
Castle station, enninton par, Cramberwell reen par have an eceptionally hih rate
of urban compleity.
uantitatively, we observe how the results obtained from the rid method represent more
uniform values with a variation between . and ., whereas values obtained from C
method vary between . and .. owever, the unconventionally hih values happen in
specific locations, which will be later discussed in the iscussion section.
ualitatively, the I method provides uniform values by proimity to the location,
while the C method presents the value of each sement independently of the nearest ones.
In this way, the I method calculation colored spots with the same compleity value are
observed due to the superficial nature of the calculation at oriin, while in the C method
larer discontinuities and differences between neihborin streets are observed since each
sement is analyed independently.
ISCSSI
oth calculation methods display similar overall results. owever, due to the different nature
of them onedimension vs twodimensions each of them contains certain bias, which are
described below
. rid ias data loss
C method avoids all data loss from calculation. here are no venue points lost from the case
study area, since every point is related to the closest street without tain this distance into
account. n the contrary, rid calculation method misses all points located in areas without
any street sement capable of representin the value of the rid. An eample would be a par
with an area over meters surrounded by streets. ven if there were any venues inside
the par, it would not be measured by any rid that at the same time includes a street sement,
so the par would appear as empty and the streets would represent a lower compleity value.
. C ias edes and culdesacs
C method produces a bias in which some small sements collect a lare amount of venues.
hose sements have almost no lenth and they are enerally plotted as edes or cul-de-sacs.
In these cases, these small sements would collect an eceptionally hih amount of venues
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
Sustainable Development and Planning IX 375
iure Calculation of Shannoniener inde per street sement usin rid ethod.
(Self-elaborated.)
iure Calculation of Shannoniener inde per street sement usin C ethod. (Self-
elaborated.)
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
376 Sustainable Development and Planning IX
because of the way in which the eometry of the networ is plotted. ue to their petite lenth,
these sements would maintain an eceptionally hih level of compleity after unifyin the
value.
here is no such problem with rid method, since it does not tae into account the lenth
of the street but it is essentially independent of any eometry.
CCSIS
his paper proved that compleity of urban space can be measured throuh Shannoniener
inde and represented alon every linear street sement within the street networ of a city.
oth methods tested C and I method seem relevant and offer a uniue
representation of the city. ein these two methodoloies uniue, each of them present a
different bias that the other could solve. owever, the overall story told by both results is
consistent and coincidin.
CS
aarini, . pe aea, ., he ushrooms esson Instaram as a tool to
evaluate users perception of urban transformations. Un nuovo ciclo della
pianificazione urbanistica tra tattica e strategia, omailano lanum ublisher, pp.
, .
heiri, A., arimipour, . orhani, ., Intraurban movement flow estimation
usin location based social networin data. Int. Arch. Photogramm. Remote Sens.
Spat. Inf. Sci. - ISPRS Arch., 40, pp. , .
Silva, .., a e elo, .., Almeida, .. oureiro, A.A., arescale study
of city dynamics and urban social behavior usin participatory sensin. Wirel.
Commun. IEEE, 21, pp. , ebruary .
olascoCirueda, A. arca ayor, C., Aplicacin de los indicadores de
compleidad urbana a travs de las redes sociales y I l caso de los paseos
martimos de evante y oniente en enidorm. XVI Congreso Nacional de
Tecnologías de la Información Geográfica. Tecnologías de la información para
nuevas formas de ver el territorio, , pp. .
oardman, ., he value of oole arth for erosion mappin. Catena, 143, pp.
, .
e ries, S., uis, A.., aners, ., aron, ., an insber, A. Sitsma, ..,
easurin the attractiveness of utch landscapes Identifyin national hotspots of
hihly valued places usin oole aps, Appl. Geogr., vol. , pp. , .
ulihe, . upia, ., appin spatial patterns of urban ariculture in ome
Italy usin oole arth and webmappin services. Land use policy, 59, pp.
, .
ustos ernndes, ., Anlisis de la compleidad urbana en la ciudad turstica l
caso de a ineda ilaseca , arraona. Int. Conf. Reg. sicence Innov. Geogr.
spillovers New approaches Evid., .
oms Castell, .., La clase creativa y la complejidad urbana de actividades en
Barcelona. Un análisis a través de datos de redes sociales dissertation niversidad
de Alicante, .
Saer, . vans, ., veryday life and locative play an eploration of oursuare
and playful enaements with space and place. Media, Cult. Soc., .
omninos, A., Stefanis, ., lessas, A. esharat, ., Capturin urban dynamics
with scarce checin data. IEEE Pervasive Comput., 12, pp. , .
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
Sustainable Development and Planning IX 377
retel, . oo, .., se and Impact of nline ravel eviews. Inf. Commun.
Technol. Tour., pp. , .
rith, ., Communicatin throuh location the understood meanin of the oursuare
checin. J. Comput. Commun., 19, pp. , .
in, ., han, ., u, . in, .., owards understandin the amification upon
users scores in a locationbased social networ. Multimed. Tools Appl., 75, pp.
, .
Schwart, . aleoua, .., he spatial self ocationbased identity
performance on social media. New Media Soc., Apr. . .
arhuus, ., rown, ., ell, ., Sherwood, S., all, . Chalmers, ., rom
awareness to repartee sharin location within social roups. In proceedings of the
SIGCHI conference on human factors in computing systems, pp. , .
oic, . euser, S., ocation based social networs definition, current state of
the art and research aenda. Trans. GIS, 17, pp. , .
Aryov, ., Alvare, ., Serranostrada, ., ortosa, . icent, .., ifferent
odels to isualise eolocated City ata from Social etwors. In Proceedings -
2015 International Conference on Future Internet of Things and Cloud, FiCloud 2015
and 2015 International Conference on Open and Big Data, OBD 2015, pp. ,
.
illier, ., Space is the machine: a configurational theory of architecture. ondon
Space Synta, .
Sevtsu, A., “Path and place: a study of urban geometry and retail activity in
Cambridge and Somerville, MA [doctoral dissertation, assachusetts Institute of
echnoloy, .
ehl, ., Svarre, ., ress, I. Steenhard, .A., How to study public life, .
Cerrone, ., ehtovouri, . pe aea, ., “The urban activity wheel. Using social
media data to map activity patterns, .
ui Snche, ., Sistemas urbanos compleos accin y comunicacin. Cuad.
Investig. urbanística, 32, pp. , .
lanco, I. Subirats, ., isten territorios socialmente ecluyentes Contra lo
ineorable. In Respuestas locales a inseguridades globales. Innovación y cambios en
Brasil y España, .
aenaar, ., overnance, compleity, and democratic participation how citiens
and public officials harness the compleities of neihborhood decline. Am. Rev.
Public Adm., 37, pp. , .
ueda, S., Ecología Urbana. arcelona eta ditorial, .
eossi, A., se of ecoloical methods in ethnobotany diversity indices. Econ. Bot.,
50, pp. , .
lorida, ., he rise of the creative class. Wash. Mon., 34, pp. , .
www.witpress.com, ISSN 1746-448X (on-line)
WIT Transactions on Ecology and The Environment, Vol 226, ©2017 WIT Press
378 Sustainable Development and Planning IX
