Revealing spatiotemporal urban activity patterns: A machine learning study using Google Popular Times

Academic Edi o : Wol gang Kainz
Recei ed: 20 Ma ch 2025
Re ised: 9 May 2025
Accep ed: 29 May 2025
Published: 3 June 2025
Ci a ion: Ba ena-He án, M.;
Mod ego-Mon o e, I.; G ijalba, O.
Re ealing Spa io empo al U ban
Ac i i y Pa e ns: A Machine Lea ning
S udy Using Google Popula Times.
ISPRS In . J. Geo-In . 2025,14, 221.
h ps://doi.o g/10.3390/ijgi14060221
Copy igh : © 2025 by he au ho s.
Published by MDPI on behal o he
In e na ional Socie y o
Pho og amme y and Remo e Sensing.
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license
(h ps://c ea i ecommons.o g/
licenses/by/4.0/).
A icle
Re ealing Spa io empo al U ban Ac i i y Pa e ns: A Machine
Lea ning S udy Using Google Popula Times
Mikel Ba ena-He án * , I zia Mod ego-Mon o e and Ola z G ijalba
CAVIAR (Quali y o Li e in A chi ec u e) Resea ch G oup, Depa men o A chi ec u e, Uni e si y o he Basque
Coun y UPV/EHU, Plaza Oña i 2, 20018 Donos ia-San Sebas ián, Spain; i zia [email p o ec ed] (I.M.-M.);
[email p o ec ed] (O.G.)
*Co espondence: mikel.ba [email p o ec ed]
Abs ac : Ex ensi e scien i ic e idence unde sco es he impo ance o iden i ying spa io em-
po al pa e ns o in es iga ing u ban dynamics. The ecen p oli e a ion o loca ion-based
social ne wo ks (LBSNs) acili a es he measu emen o u ban hy hms h ough geo empo-
al in o ma ion, p o iding deepe insigh s in o he unde lying causes o u ban ib ancy.
This s udy p esen s a me hodology o analyzing he spa io empo al use o ci ies and
iden i ying occupancy pa e ns aking in o conside a ion u ban o m and unc ion. The
analysis elies on da a ob ained om Google Popula Times (GPT), ans o ming he ela-
i e occupancy o a la ge numbe o poin s o in e es (POI) classi ied in o i e ca ego ies,
o es ima ing he numbe o people agg ega ed wi hin u ban nodes du ing a ypical day.
As a esul , his esea ch assesses he u ili y o his da a sou ce o e alua ing he changing
dynamics o a ci y ac oss bo h space and ime. The me hodology employs geog aphic in-
o ma ion sys em (GIS) ools and a i icial in elligence echniques. The esul s demons a e
ha by analyzing geo empo al da a, we can classi y u ban nodes acco ding o hei hou ly
ac i i y pa e ns. These pa e ns, in u n, ela e o ci y o m and u ban ac i i ies, showing a
ce ain spa ial concen a ion. This esea ch con ibu es o he g owing body o knowledge
on machine lea ning (ML) me hods o spa io empo al modeling, laying he g oundwo k
o u u e s udies ha can u he explo e he complexi y o u ban phenomena.
Keywo ds: spa io- empo al analysis; ime se ies clus e ing; u ban dynamics; loca ion-based
social ne wo k; geog aphic in o ma ion sys ems; ime geog aphy
1. In oduc ion
1.1. Gene al O e iew
The ela ionship be ween he spa ial cha ac e is ics o a ci y and human beha io
has been s udied o many decades. T adi ionally, he scien i ic app oach has been sec-
o al: mo phological o u ban s udies ha e been conduc ed wi hin he mo e echnical o
a chi ec u al ields, while analyses o human beha io and dynamics adi ionally o ig-
ina ed om humanis ic pe spec i es such as sociology and psychology. Howe e , he
con inuous ad ancemen o geog aphic in o ma ion sys em (GIS) ools, he p oli e a ion o
geoloca ed da a, and eme ging a i icial in elligence (AI) echniques a e opening up a new
ield o esea ch.
In his con ex , loca ion-based social ne wo ks (LBSNs) p o ide in o ma ion abou
human ac i i y wi hin geo- e e enced digi al en i onmen s. Speci ically, Google Popula
Times (GPT), whose spa io empo al da a a e openly accessible in eal ime and a ailable
globally, o e s ex ensi e geoloca ed da a om poin s o in e es (POI) ha cap u e he
ISPRS In . J. Geo-In . 2025,14, 221 h ps://doi.o g/10.3390/ijgi14060221
ISPRS In . J. Geo-In . 2025,14, 221 2 o 23
occupancy le els o es ablishmen s and public spaces. Howe e , while he use o his sou ce
in u ban s udies has seen conside able p og ess, signi ican me hodological ad ancemen s
a e s ill equi ed o ensu e ha he esul s a e bo h ep esen a i e and alid o s a egic
u ban design.
In his ega d, machine lea ning (ML) eme ges as an indispensable ool. I de el-
ops algo i hms and s a is ical models ha allow compu e sys ems o lea n om and
p ocess la ge olumes o da a, enabling a mo e comp ehensi e unde s anding o human
ac i i y pa e ns.
The e o e, new echniques associa ed wi h he p oli e a ion o LBSNs, combined wi h
he enhanced capabili ies o AI in big da a analysis, a e d i ing signi ican ad ancemen s
in he me hodological de elopmen o spa io empo al analysis o u ban li e. These new
me hods will os e a deepe unde s anding o u ban dynamics and acili a e mo e in o med
decision making in u ban planning.
This pape in oduces an inno a i e me hodology o he spa io empo al analysis o
occupancy pa e ns, in eg a ing he unique u ban o m and he a ious ac i i ies ha un old
wi hin ine-g ained u ban uni s o analysis. By employing his app oach, we can iden i y
u ban hy hms h oughou ime pe iods o each u ban node, allowing us o e alua e he
dynamics o a eas wi h a ying le els o u ban ib ancy and ul ima ely in o m s a egic u ban
planning decisions. The me hodology is applied o he case s udy o Donos ia-San Sebas ián.
1.2. Li e a u e Re iew
The s udy o human daily ac i i y ou ines h ough space and ime in u ban se ings
began in he la e 1960s, leading o he de elopmen o a new ield o esea ch: ime geog a-
phy. This made i possible o analyze human beha io al pa e ns in eg a ing bo h spa ial
and empo al dimensions hough spa io empo al p isms [
1
]. O he classical app oaches
in sociology and human geog aphy, such as ime-use su eys and spa io empo al dia ies,
allowed u ban planne s in he 1970s o be e unde s and he unc ioning o ci ies and
e eal socio-spa ial dynamics by in oducing he a iable o ime. As such, he u ban ab ic
is unde s ood as a con ex o beha io [
2
] in luencing la ge-scale ime o ganiza ion and
he hy hms o human ac i i y in e e yday li e [
3
]. In his con ex , he Theo y o Na u al
Mo emen and Space Syn ax u he emphasizes how he con igu a ion o he u ban g id
p i ileges ce ain spaces o h ough mo emen [
4
], he eby shaping hese socio-spa ial dy-
namics and ein o cing he hy hms o daily ac i i y. Ci ies a e hus composed o mul iple
ime ames and luid empo ali ies o e en s [5].
F om an ecological pe spec i e, cha ac e izing he daily lows o di e en communi ies
wi hin a ci y’s spa ial s uc u e [
6
,
7
] led o he conclusion ha simila u ban con ex s
o en sha e empo al egula i ies [
8
]. Fo ins ance, Goodchild’s spa io empo al analyses,
combining social da a and ca og aphy, showed ha he p ima y o ganizing dimension
o u ban space is he ela ionship be ween esidence and wo k [
9
], while socio-economic
s a us is also a c i ical a iable, in luencing he ime spen on a ious ac i i ies [10].
Addi ionally, ac i i y censuses in la ge ci ies demons a e hy hms o social beha -
io [
11
], akin o he “mechanical pe iods” o social li e, shaped no only by he u ban con ex
bu also by sha ed wo k schedules, habi s, and adi ions ha c ea e a hy hm and sense o
place. Rou ine human ac i i ies o en ake place in p oxima e, small spaces and a e linked
o sho pe iods o ime [
12
]. Consequen ly, he ele ance o a place is ied o i s empo al
hy hm, and ice e sa; ce ain hy hms end o spa ialize [13].
O e he las decade, a new pa adigm has eme ged in which u ban hy hm da a
a e linked o loca ions a he han indi iduals [
14
], highligh ing a signi ican associa ion
be ween buil en i onmen ac o s and u ban ib ancy [
15
]. This makes i possible o
iden i y he mo emen , mee ing, and es ing pa e ns o people in ci ies, gi ing each
ISPRS In . J. Geo-In . 2025,14, 221 3 o 23
place a dis inc i e, inhe en ly hy hmic cha ac e [
16
]. Fo ins ance, by combining Jane
Jacobs’ pa ame e s o u ban i ali y [
17
] wi h obse a ions o pedes ian lows a u ban
in e sec ions, poly hy hms—pa e ns c ea ed by ou ines, social in e ac ions, and anspo
sys ems—can be de ec ed [
18
]. Thus, each ac i i y ch ono ype [
19
] es ablishes a empo al
connec ion be ween spa ially sepa a ed places [20].
In his con ex , nume ous s udies ha e analyzed ci ies om his spa io empo al di-
mension, linking he cha ac e is ics o he buil en i onmen o he dynamics gene a ed
wi hin i . Fo ins ance, li es yle changes b ough abou by globaliza ion and inc eased ime
use demands [
21
] mani es as ex eme u ban i ali y and socio-cul u al iden i y in 24/7 ci y
cen e s. These en i onmen s ex end beyond adi ional wo king hou s, e ealing mul iple
di isions in isi a ion in ensi y and ac i i ies [
22
], leading o bo h spa ial and empo al
seg ega ion [
23
]. This lack o inclusi i y s ems om he limi ed a ie y o se ices ha mee
he needs o di e en social g oups [
24
]. Addi ionally, he ise in he nigh ime economy
ans o ms ci ies in o spaces o s anda dized consump ion, a ac ing homogeneous g oups
o consume s, ou is s, and en ep eneu s [
25
], which esul s in challenges o esiden s
as gen i ica ion and he exclusion o ce ain social g oups [
26
]. In esponse, u ban s a e-
gies a e eme ging ha seek o in eg a e and manage he day ime, e ening, and nigh ime
economies based on cus ome expe ience and pe cep ion in ci y cen e s [27].
Mo eo e , he exposu e o people o he e ogeneous social con ex s depends on hei
indi idual cha ac e is ics and he ac i i y spaces hey equen [
28
]. Obse ing hese spaces
helps iden i y beha io al pa e ns, he physical cha ac e is ics o s ee s ha encou age s a-
iona y and pe sis en ac i i ies [
29
], as neighbo hood cha ac e and dynamics a e shaped
by u ban o m [
30
], and ac i i y dis ibu ions ha e lec socio-spa ial cha ac e is ics [
31
].
E en in he mos in ima e spaces, like s ee s and squa es, he e is a di e si y and luidi y
in he encoun e s and mo emen s ha cons i u e he u ban ab ic [32].
U ban dynamics, as e idenced h ough le els o ac i i y, comme cial a ailabili y, and
household con ibu ions o he economy, shape bo h public li e and he socio-spa ial o ga-
niza ion o ci ies [
33
]. The scale and size also play a c ucial ole, as pa e ns o socializa ion
di e by u ban densi y, wi h s onge connec ions in compac ci ies, due o physical and
design ac o s like public spaces, building ypologies, and isual spaciousness [
34
]. The
quali a i e expe ience o u ban en i onmen s can hus be desc ibed h ough empo al,
spa ial, isual, and connec i i y me ics o u ban o m [35].
Wi h he ad ancemen o in o ma ion echnologies and he inc easing spa io empo al
esolu ion o da a, in eg a ing space and ime in o GIS en i onmen s p esen s a challenge,
especially when ying o isualize a ci y’s mul i- unc ionali y [
36
]. In e es ingly, hese ech-
nologies ha e al e ed he iming and loca ion o ac i i ies [
37
], eshaping socio-economic
pa e ns in u ban a eas. Compu a ional imp o emen s and GIS ad ancemen s [
38
] ha e
played a c ucial ole in anspo planning [
39
] and ha e e i ed he s udy o socio-u ban
dynamics [40].
Recen esea ch on u ban dynamics has inc easingly adop ed spa io empo al pe spec-
i es and big da a sou ces, e lec ing a g owing in e es in unde s anding he complexi y
o ci y li e h ough digi al aces [
41
], whe e ML enhances spa ial analysis by iden i ying
complex pa e ns, in eg a ing di e se da a sou ces, and adap ing dynamically o u ban
challenges.
A wide a ie y o da a sou ces ha e been explo ed, including mobile phone eco ds
[
42
–
45
], social media con en [
46
–
49
], ansi sma ca d da a [
50
,
51
], bike sha ing
sys ems [
52
–
54
], axi GPS ajec o ies [
55
–
57
], and inancial ansac ions [
58
]. These s udies
add ess di e se hema ic angles—such as u ban ib ancy [
15
,
53
,
59
], unc ional land use
and POIs de ec ion [
46
–
48
,
60
], commu ing and mobili y pa e ns [
50
,
51
,
61
], conges ion anal-
ISPRS In . J. Geo-In . 2025,14, 221 4 o 23
ysis [
55
,
56
], and esilience o dis up ions [
62
]—demons a ing he ichness and po en ial o
spa io empo al u ban esea ch.
In e ms o analy ical echniques, ecen con ibu ions showcase an e ol ing me hod-
ological landscape. Clus e ing me hods emain cen al, wi h k-means equen ly used o
unco e land use pa e ns o mobili y dynamics [
48
,
51
,
53
,
57
], while o he s udies adop
mo e sophis ica ed app oaches such as Dynamic Time Wa ping (DTW) combined wi h
k-medoids o delinea e unc ional zones based on building-le el social media ac i i y [
46
],
o hie a chical DTW o iden i y cyclical beha io al pa e ns in bicycle usage [
52
]. Modi-
ied DBSCAN and uzzy clus e ing algo i hms ha e been employed o de ec commu ing
lows and conges ion dynamics wi h g ea e empo al nuance [
50
,
54
,
55
]. Neu al ne wo ks
ha e also been in eg a ed o hou ly popula ion densi y es ima ion [
48
] and i ali y a ea
classi ica ion [
59
], while g a i y-based models [
42
] and spa io empo al low clus e ing
s a egies [61] ha e eme ged o cap u e in e ac ion in ensi ies and mobili y ends.
Beyond echnique, he hema ic ocus on u ban hy hms is pa icula ly ele an .
Mul iple s udies examine in a-u ban a ia ions in ac i i y in ensi y, empo ali y, and
unc ion—highligh ing he co-dependence be ween land use, mobili y, and buil en i on-
men s uc u es [
15
,
59
,
60
,
63
]. O he s explo e he esilience o u ban sys ems unde ex e nal
shocks, such as ex eme wea he , by le e aging empo ally ich da ase s like GPT o de ec
shi s in daily ou ines [62].
Despi e hese ad ances, impo an gaps emain. Fi s , ew s udies ely on publicly
accessible and globally consis en da ase s, such as GPT, which o e scalable and eplicable
insigh s in o human ac i i y while a oiding many p i acy issues inhe en o mobile o
inancial da a. Second, while ad anced clus e ing and modeling echniques a e widesp ead,
he e is a lack o me hodological s anda diza ion, which hinde s compa a i e analysis
ac oss ci ies o egions. Thi d, ela i ely ew con ibu ions o e in eg a ed spa ial and
empo al g anula i y, which is essen ial o unde s and he ine-scale hy hms o u ban li e,
pa icula ly a he in a-neighbo hood le el.
In his con ex , he p esen s udy con ibu es a eplicable and ligh weigh me hodol-
ogy ha le e ages GPT and unsupe ised ML echniques o classi y occupancy pa e ns
ac oss u ban space and ime. By ocusing on unc ional u ban hy hms and hei spa ial
mani es a ion, i add esses bo h he me hodological agmen a ion and da a accessibili y
limi a ions iden i ied in p io wo k.
2. Ma e ials and Me hods
2.1. S udy A ea
In pu sui o applying he me hodology o a local con ex , he selec ed case s udy o
alida ing he me hod is he municipali y o Donos ia-San Sebas ián (Figu e 1), a coas al
ci y in no he n Spain.
The ci y o igina ed as a walled medie al se lemen , and i s g ow h o e ime ex ended
ac oss he we lands o he U umea Ri e and along he coas al edge. Today, i p esen s
a clea mo phological s a i ica ion: he his o ical medie al co e, he 19 h-cen u y and
pos mode n u ban expansions ha cha ac e ize he lowland dis ic s, and mo e ecen
pe iphe al neighbo hoods and subu bs loca ed on su ounding hillsides and sloped e ain.
Despi e he ci y’s mode a e size—174,529 inhabi an s wi hin he u ban a ea [
64
]—i s
high popula ion densi y o 13,073 inhabi an s/km
2
exhibi s a “Medi e anean” li es yle.
This, along wi h i s unc ional cha ac e is ics, e lec s sociospa ial dynamics whe e balanc-
ing wo k and li e can be challenging [65].
ISPRS In . J. Geo-In . 2025,14, 221 5 o 23
Figu e 1. The adminis a i e di ision o Donos ia-San Sebas ián. Sou ce: GeoEuskadi. Own elabo a ion.
As o 2022, in he Basque Coun y, 99% o indi iduals aged 16 o 74 egula ly use
sma phones [
66
]. Addi ionally, 56.5% o in e ne use s engage wi h social ne wo ks, and
55.2% o companies use social ne wo ks o business pu poses [
67
]. This widesp ead
adop ion suppo s he po en ial o LBSNs o u ban s udies in his con ex .
S ee -le el u ban uses in Donos ia-San Sebas ián consis o a a ie y o ac i i ies wi h
di e ing ope a ing hou s h oughou he day [
68
]. The majo i y o hese businesses include
e ail, hospi ali y, and auxilia y p o essional se ices, ollowed by public se ices. This
comme cial di e si y is highly concen a ed in he dense neighbo hoods loca ed in he
la e a eas o he ci y, making i a p ime a ea o s udying u ban dynamics.
2.2. Me hodology
To add ess his s udy’s objec i e o iden i ying and classi ying unc ional u ban
hy hms, a combina ion o empo al, spa ial, and mo phological dimensions was equi ed.
The i s s ep in ol ed agg ega ing POIs wi hin a mo phological g id ha cap u es he
physical s uc u e o he ci y. Unlike con en ional adminis a i e bounda ies, he use o
mo phologically homogeneous uni s allows o a mo e consis en compa ison o ac i i y
ac oss u ban space, aligning wi h p e ious esea ch ha emphasizes he impo ance o he
buil en i onmen in shaping beha io . Mo eo e , weigh ing each uni by he legal capaci y
o i s POIs enables he iden i ica ion o a eas no simply by he coun o es ablishmen s, bu
by hei po en ial in ensi y o use.
Fo empo al clus e ing, we selec ed he k-shape algo i hm, which is speci ically
designed o no malized ime se ies. Unlike classical clus e ing echniques such as k-
means, k-shape accoun s o bo h he shape and alignmen o empo al pa e ns, making
i pa icula ly sui able o cap u ing cha ac e is ic ac i i y p o iles while being obus
o di e ences in ampli ude. This was c ucial o dis inguish usage pa e ns ha ollow
simila hy hms e en i hei absolu e magni udes di e . This ype o unsupe ised ML is
pa icula ly use ul in e ealing la en spa io empo al pa e ns ha a e no easily cap u ed

ISPRS In . J. Geo-In . 2025,14, 221 6 o 23
by con en ional GIS o s a is ical echniques, allowing o a mo e nuanced and da a-d i en
classi ica ion o u ban ac i i y p o iles.
Finally, hea maps we e used o explo e he spa ial dis ibu ion o he esul ing clus e s.
This spa ial smoo hing echnique highligh s concen a ions and dispe sions o ac i i y
ypes ac oss he ci y, o e ing an in ui i e isualiza ion o unc ional zones and hei
mo phological con ex . This in eg a i e app oach suppo s bo h analy ical igo and
spa ial in e p e abili y, laying he g oundwo k o linking usage pa e ns o planning
conside a ions.
Figu e 2illus a es he i e s ages in which he me hodology is s uc u ed: (a) collec ion
o aw da a and p ocessing, (b) de ini ion o he uni o analysis, (c) agg ega ion o capaci y-
weigh ed POIs occupancy a es, (d) applica ion o ime se ies clus e ing echniques, and (e)
spa ial analysis o he occupancy ends.
Figu e 2. Me hod low diag am.
This app oach in eg a ed wo ypes o spa ial da a: (1) an LBSN, (2) he adminis a i e
egis y o u ban eal es a e.
Rega ding LBSNs, Google Places is cu en ly one o he mos comp ehensi e global
da abases o u ban s udies, o e ing aluable insigh s in o land use a ibu es and u ban
dynamics. Howe e , Google Places lacks empo al da a, as POIs a e s a ic in na u e. The
in oduc ion o GPT has signi ican ly changed his scena io. GPT calcula es occupancy
pa e ns by analyzing isi da a, agg ega ed and anonymized, collec ed o e he p e ious
ou o six weeks. The peak hou is used as he e e ence, wi h o he es ima es displayed in
ela ion o his peak [
69
]. GPT has hus become an inc easingly popula ool in esea ch
ela ed o u ban dynamics.
(a)
Collec ion and P ocessing o Ca og aphic and GPT Da a
To collec and p ocess he necessa y POIs a ibu es, we employed he c awle -google-
places ool a ailable on Gi Hub, using commi 12c124 [
70
]. Da a ex ac ion was conduc ed
ia an uno icial API, which au oma ed he e ie al o POIs a ibu es by en e ing links ob-
ained h ough a combina ion o selec ed Google Places ca ego ies and desi ed geog aphic
loca ions. The da ase includes a a ie y o ields ela ed o each POI, whose key a ibu es
o in e es —ob ained a e a se ies o p ep ocessing s eps o o ganize he da a in o he
co esponding ields—a e:
ISPRS In . J. Geo-In . 2025,14, 221 7 o 23
-
Popula imes: Hou ly a e age occupancy as a pe cen age ela i e o he peak occu-
pancy.
- Ca ego y: The main ca ego y decla ed by he owne on Google My Business.
- Geoloca ion: Geog aphical coo dina es.
- Add ess: S ee name and doo numbe .
Fo his s udy, da a we e collec ed om 1378 POIs loca ed in Donos ia-San Sebas ián on
a F iday du ing Ap il and May 2022, allowing us o cap u e bo h ou ine and leisu e-d i en
u ban dynamics wi hin a single day
Figu e 3shows a b anched diag am ha ca ego izes he sea ches ca ied ou using
Google Places, g ouped in o i e main classes. This classi ica ion—ba s and es au an s,
shops, wellbeing, p o essional se ices, and ou doo s—was de ined o syn hesize he
mul i ude o highly speci ic ca ego ies p o ided by Google Places in o b oade unc ional
g oups, allowing o a mo e cohe en and in e p e able analysis o u ban ac i i y. The
diag am also displays he numbe o POIs wi hin each class and hei pe cen age ela i e
o he o al.
Figu e 3. Classi ica ion o POIs in he municipali y o Donos ia-San Sebas ián. The size o he symbols
is p opo ional o he numbe o POIs. Sou ce: Google Places. Own elabo a ion.
Addi ionally, we collec ed he cadas al da abase o u ban p ope ies in he municipal-
i y [
71
], which uniquely iden i ies each p ope y and p o ides de ailed u ban cha ac e is ics,
including add ess, buil -up a ea, and he in ended use o each p ope y.
A e bo h da ase s we e p epa ed, hey we e linked h ough hei s anda dized
add esses o associa e he cadas al loo a ea wi h he POIs da a. In cases whe e no di ec
ma ch exis ed be ween he wo da ase s, he POIs we e assigned he a e age loo a ea o
ISPRS In . J. Geo-In . 2025,14, 221 8 o 23
i s espec i e sub-ca ego y o ensu e consis ency in he analysis and a oid missing alues
ha could impac he in e p e a ion o spa ial dis ibu ion pa e ns.
To es ima e he numbe o people occupying each POI, we e e enced he Spanish
Technical Building Code (CTE), pa icula ly he Basic Documen on Fi e Sa e y (CTE
DB SI). This egula ion de ines he maximum occupancy densi y o di e en p ope y
ypes, measu ed in squa e me e s pe pe son. Using his s anda d, each POI is assigned
a maximum occupancy alue (capaci y), which se es as a weigh ing ac o . This allows
us o ans o m he ela i e GPT occupancy alues in o an absolu e numbe o es ima ed
people pe hou .
(b)
Uni o Analysis: he Mo phological G id
The uni o analysis o his esea ch, he elabo a ion s eps o which a e shown in
Figu e 4, is based on a g id sys em adap ed o he ci y’s u ban mo phology. This i egula
g id was gene a ed using Vo onoi polygons o igina ing om s ee in e sec ions, which a e
he s a egic poin s o connec ion and decision o people in mo ion [
72
]. Thus, such clea
isual join s p o ide a ine-g ained delimi a ion o unc ional nodes, main aining ela i e
spa ial homogenei y a he han elying on a delimi a ion based on u ban blocks, which
would conside only a single acade o each s ee hey encompass.
Figu e 4. Design p ocess o he mo phological g id om he s ee ne wo k. (1) The o iginal s ee
ne wo k includes all line ea u es mapped o each s ee (e.g., bike lanes, a ic di ec ions); (2) he
simpli ied s ee ne wo k educes each s ee o a single cen al axis; (3) in e sec ions a e gene a ed
a he c ossing poin s o hese axes, shown as ‘+’ symbols; (4) Vo onoi polygons a e c ea ed om
he in e sec ions o de ine he mo phological g id, aligning wi h he u ban s uc u e. Colo s a e
andomly assigned o aid in e p e a ion. Sou ce: GeoEuskadi. Own elabo a ion.
(c)
Weigh ed Occupancy Calcula ion
The POIs a e spa ially linked o he co esponding cells wi hin he mo phological
g id. As mul iple POIs may coexis wi hin a single cell, hei weigh ed con ibu ion o
he hou ly occupancy calcula ion is de e mined by hei ela i e capaci y in p opo ion
o he o al capaci y o all POIs wi hin ha cell. To es ima e he capaci y o each POI, a
ca ego y-dependen occupancy densi y a io (Table 1) was applied o i s cadas al loo a ea.
Pj=
n
∑
i=1Ai
δc,i(1)
P= capaci y (maximum numbe o people);
A= loo a ea (m2);
δ= densi y (m2/pe son);
ISPRS In . J. Geo-In . 2025,14, 221 9 o 23
j= cell;
i= POIs;
c= ca ego y.
Table 1. Ra io o he numbe o occupan s o he loo a ea o a habi able uni . Sou ce: Código Técnico
de la Edi icación. Documen o Básico Segu idad en caso de Incendio (CTE DB SI). Own elabo a ion.
In ended Use Acco ding o CTE Occupa ion
(m2/Pe son) POI Ca ego y
Adminis a i e 10
Lawye
Ad e ising agency
A chi ec
Bank
Managemen
Company o ices
Comme cial
2
Bu che ’s shop
Beau y salon
Clo hes shop
G oce y
Bake y
Hai d esse
Fishmonge s
Pha macy
G eeng oce ’s
3
Copy shop
Cou ie se ice
Compu e shop
5 Supe ma ke
Public
1 Ba s
0.5
Pub
Dance club
Disco club
1.5 Res au an
5 Gym
10 Squa e
Tou is a ac ion
25 1Pa k
Hospi al
10 Den is
Nu i ionis
15 Gynaecologis
Physio he apis
1Acco ding o u ban s anda ds o open spaces (pa ks) in a medium ab ic a neighbo hood-ci y scale.
Subsequen ly, he hou ly calcula ion o weigh ed occupancy was de e mined by
combining he ela i e weigh o each POI wi hin he cell and i s co esponding occupancy
alue om GPT. Fo each cell and hou ly in e al, i was calcula ed as ollows:
O ,j=
n
∑
i=1 Pi
Pj
∗GPT ,i!(2)
O= weigh ed occupancy a e (%);
ISPRS In . J. Geo-In . 2025,14, 221 16 o 23
e al a eas, he dis ibu ion di e ges, o ming smalle , mo e isola ed concen a ions wi h
lowe in ensi y.
Finally, C5 is s ongly localized a ound emblema ic squa es wi hin he ci y cen e ,
pa icula ly he Ca hed al Squa e, as well as key plazas in he Pa e Vieja and An iguo.
This clus e is i ually absen in pe iphe al neighbo hoods, ein o cing i s associa ion wi h
cen ali y and symbolic u ban spaces.

ISPRS In . J. Geo-In . 2025,14, 221 17 o 23
Figu e 9. The 5 clus e s in Donos ia-San Sebas ián and hei espec i e (a) ke nel densi y maps
highligh ing he mos ep esen a i e a eas, (b) ada cha s o mixed-use composi ion by POIs
ca ego ies, and (c) s acked a ea g aphs o empo al occupancy by POIs ca ego ies. Each clus e is
ep esen ed by a speci ic colo . The i e ca ego ies a e colo -coded and iden i ied in he legend below.
4. Discussion
The ML-based me hodology applied in his s udy enabled he classi ica ion o u ban
space in o i e dis inc empo al clus e s, each e lec ing cha ac e is ic occupancy pa e ns
de i ed om GPT da a. These clus e s e eal bo h daily hy hms o ac i i y and hei
ISPRS In . J. Geo-In . 2025,14, 221 18 o 23
spa ial dis ibu ion, highligh ing he di e en ia ed oles ha neighbo hoods play wi hin
he unc ional s uc u e o he ci y.
Al hough all clus e ypes a e ep esen ed ac oss he ci y’s neighbo hoods—consis en
wi h he mul i unc ional cha ac e o a compac , mixed-use, and complex ci y such as
Donos ia-San Sebas ián— he KDE e eals dis inc spa ial dis ibu ions o each unc ional
clus e . This allows o he associa ion o occupancy dynamics wi h he ci y’s mo pho-
logical and s uc u al cha ac e is ics. As shown in he esul s, his case s udy enables he
iden i ica ion o key axes o ho spo s o each occupancy pa e n.
O e all, he cen al pa o he ci y and he adjacen neighbo hoods—mainly loca ed
in la a eas wi h mo e homogeneous u ban s uc u es—hos highe le els o ac i i y ac oss
all clus e ypes. In con as , mo e pe iphe al and opog aphically ele a ed a eas show
signi ican ly lowe ac i i y le els, wi h ce ain esiden ial neighbo hoods—de eloped
du ing he 1970s and 1980s o in mo e ecen yea s—showing an almos comple e absence
o concen a ed u ban unc ions. This con i ms a clea cen e –pe iphe y g adien in he
ci y’s po en ial o a ac people.
A a ine scale, deepe analysis wi hin indi idual neighbo hoods e eals ha he mos
in ense loca ions o each clus e end o align wi h pedes ian co ido s, public squa es, o
main anspo a ion ou es. As demons a ed in Sec ion 3.2.2, he p oposed me hod enables
he iden i ica ion o dis inc spa ial pa e ns based on bo h he ype o clus e and he
exis ing u ban ab ic. These indings could o m he basis o u u e, mo e de ailed s udies
on he ela ionship be ween occupancy pa e ns and u ban mo phology and unc ions a
he neighbo hood scale.
The empo al luc ua ions o each clus e also di e ma kedly, o e ing insigh in o he
unc ional d i e s o human agglome a ion. Fo example, ba -o ien ed clus e s (C2 and C5)
ha e he highes con as o agglome a ion a ios and show a high spa ial concen a ion.
On he one hand, C5 has he absolu e highes e ening peak bu was p eceded by eally
low mo ning ac i i y. On he o he hand, C2 shows a deep alley in he a e noon. The
mo e e enly dis ibu ed mixed-use cells ep esen ed in C1 ha e a cha ac e is ic mo ning
peak. Meanwhile, he shopping-o ien ed cells o C4 a e dispe sed h oughou he ci y,
al hough highe in ensi ies clea ly mani es in cen al a eas. I s mul i unc ional cha ac e
allows o a con inuous ac i i y pa e n, simila o C3, which ins ead shows he highes
spa ial concen a ion and a di e en o e ing o ac i i ies ha end owa d wellbeing and
ee spaces.
These pa e ns di e in iming, in ensi y, and unc ional composi ion. Clus e s wi h
p onounced e ening peaks d i en by ba s and es au an s (e.g., C5) con as wi h hose
ha show symme ical mo ning and a e noon ac i i y linked o e ail (e.g., C4). Spa ially,
hese pa e ns co espond o es ablished u ban hie a chies, echoing Ch is alle ’s cen al
place heo y by iden i ying key cen e s in he his o ic co e and s uc u ed expansion
a eas. Howe e , unlike s a ic spa ial models, ou da a-d i en app oach cap u es eme gen
unc ional hie a chies based on eal- ime usage, aligning wi h ecen heo e ical upda es
ha inco po a e empo al specializa ion in o u ban hie a chy models [20].
Ou esul s a e in line wi h ecen s udies highligh ing empo al specializa ion in
u ban en i onmen s. Fo ins ance, he di e gence be ween midday and nigh ime check in
beha io s ac oss neighbo hoods [
64
], sugges ing ime-speci ic unc ionali y independen
o ca ego ical di e si y. In ou indings, he a eas wi h s ong e ening ac i i y (e.g., C5)
a e no necessa ily he mos unc ionally di e se bu a e he mos in ense in one ca ego y
(ba s and es au an s), ea i ming ha spa ial specializa ion and empo al hy hm do no
always align wi h land use a ie y.
F om a me hodological pe spec i e, ou app oach builds on he ad ances in u ban
pa e n mining h ough ime se ies clus e ing. While p io wo k has applied his o
ISPRS In . J. Geo-In . 2025,14, 221 19 o 23
single domains—such as bike sha ing sys ems [
74
] o nigh li e eco e y [
75
]—ou mul i-
ca ego ical pe spec i e cap u es he coexis ence and laye ing o u ban unc ions, a de ining
ai o compac ci ies like Donos ia-San Sebas ián.
Ou decision o agg ega e he POIs wi hin mo phologically homogeneous u ban
uni s also imp o es spa ial g anula i y. In con as o s udies ha ely on adminis a i e
bounda ies o uni o m g ids [
15
], ou i egula mo phological g id aligns mo e closely
wi h he s ee ne wo k and buil o m. This enhances spa ial ele ance and in e p e abili y.
Mo eo e , by weigh ing he occupancy based on POIs ca ego y and cadas al su ace,
we con e GPT’s ela i e me ics in o es ima ed coun s o people, add essing a majo
sho coming o s udies using pu ely ela i e indica o s [62].
The p oposed amewo k was de eloped using open-sou ce ools, including Py hon
and QGIS o spa ial p ocessing, ensu ing anspa ency and ep oducibili y. Gi en he
ligh weigh na u e o he algo i hms and he localized scope o he analysis, he compu-
a ional cos and en i onmen al impac emain minimal compa ed wi h mo e in ensi e
AI amewo ks.
Tha said, he me hodology has limi a ions. The mul isou ce s a egy equi es ha -
monizing da a om di e en p o ide s, pa icula ly ma ching Google Places POIs wi h
cadas al p emises. This p ocess in ol es add ess s anda diza ion, and is subjec o geolo-
ca ion e o s, misma ches, o absen eco ds. In such cases, we assigned a e age ca ego y
su ace alues, which in oduces some unce ain y in o he spa ial analysis.
Es ima ing maximum occupancy also assumes ha he highes GPT- eco ded alue
e lec s he legal capaci y o each enue, which may no align wi h ac ual peak ac i i y o
compliance [
76
]. Fu he mo e, GPT’s olling a e age is sensi i e o excep ional e en s (e.g.,
holidays, closu es), po en ially skewing he baseline o no mal beha io .
We also ecognize he dependency on a p op ie a y pla o m. Al hough GPT p o ides
ine-g ained, publicly accessible da a, changes in i s access policies o da a a chi ec u e
could a ec u u e esea ch con inui y. Mo e undamen ally, Google Places and GPT
may o e ep esen comme cial ac i i y while unde ep esen ing sec o s like educa ion,
heal hca e, o in o mal uses. Despi e he widesp ead use o sma phones and LBSNs,
digi al beha io is s ill shaped by sociodemog aphic biases, meaning some popula ion
g oups a e unde ep esen ed [77].
Finally, while his s udy segmen s u ban space based on mo phological s uc u e,
i does no di ec ly quan i y u ban o m h ough a iables like block size, densi y, o
connec i i y. Fu u e wo k should build on his amewo k by in eg a ing spa ial indica o s
in o co ela ion models [
59
,
63
], o mo e clea ly a icula e how u ban o m in luences
usage dynamics.
5. Conclusions
This pape p esen s a me hod o he spa io empo al analysis o u ban dynamics,
ocusing on he concen a ion o people wi hin a ci y. I e alua es he e ec i eness o using
Google Maps and land use da a, combined wi h ML echniques o measu e ine-g ained
u ban occupancy pa e ns.
The a ailabili y o geosocial da a h ough LBSNs o e s new oppo uni ies o de ailed
s udies o he spa io empo al concen a ion o people wi hin ci ies. As he usage o hese
pla o ms g ows ac oss a ious popula ion g oups, he po en ial and ep esen a i eness o
hese da a a e expec ed o inc ease, p o ided ha public access o such da a emains a ail-
able. Howe e , p e ious s udies ha e no conside ed ei he he capaci y o he in e ac ion
be ween POIs.
So, he in eg a ion o s a ic da a and publicly accessible geo empo al da a wi h ML
ep esen s a signi ican ad ancemen in he s udy o u ban dynamics. This me hodology
ISPRS In . J. Geo-In . 2025,14, 221 20 o 23
no only s anda dizes da a analysis o people agglome a ion bu also enables scalabili y o
o he u ban con ex s.
The indings demons a e ha by analyzing geoloca ed, ime-s amped big da a, i
is possible o model he ac i i y pa e ns o speci ic u ban nodes. This me hodology
enables de ailed spa io empo al analysis, o e ing aluable insigh s in o he hou ly beha io
o people in a ci y in ela ion o u ban o m and i s unc ions ep esen ed by spa ially
ela ed POIs. The es ima ion o he olume o people, using ca ego y-dependen densi y
a ios and he loo a ea o POIs, weigh ed by GPT da a allows o an hou ly capaci y
calcula ion o each loca ion. This app oach in ol es a e e se-enginee ing p ocess using
pe cen age alues p o ided by Google om agg ega ed and anonymized da a, all o which
is publicly accessible.
Fu he mo e, his s udy con i ms ha he ela ionship be ween land use and ac i i y
pa e ns emains a c i ical ac o in u ban li e. The case s udy also highligh s he impo ance
o unde s anding he spa ial in e ac ion o di e en POIs ca ego ies in shaping u ban
dynamics h oughou he day. The insigh s gained om clus e ing spa io empo al ac i i y
pa e ns p o ide a amewo k o a ge ed u ban planning in e en ions, as he obse ed
empo al hy hms in a eas wi h mixed land use ein o ce he ole o u ban o m and
unc ion in shaping ac i i y pa e ns.
I also allows o he iden i ica ion o cha ac e is ic u ban pa e ns o hy hms by in-
co po a ing ML echniques in o he analysis o u ban big da a. Fu he mo e, he clus e ing
o ime se ies da a o e s a clea isual ep esen a ion o u ban a eas ha sha e simila
unc ionali y and ac i i y pa e ns, e en when hose a eas a e geog aphically discon inuous.
By in eg a ing empo al and spa ial da a on he p esence o people in a ci y, a new
dimension is in oduced o he complex analysis o u ban dynamics. In conjunc ion
wi h he s udy o u ban s uc u e and o m, his app oach enables he iden i ica ion o
imbalances occu ing wi hin a ci y, as well as he cha ac e iza ion o la ge dep essed o low-
i ali y a eas. Addi ionally, i allows o he de ec ion o zones wi h excessi e occupancy,
which may be conside ed sa u a ed.
The e o e, he p oposed me hodology in oduces a no el app oach ha suppo s
da a-d i en u ban planning decisions, which could be applied in he e alua ion and design
o municipal policies in a ious a eas such as mobili y, he de elopmen o local economies,
o he loca ion o public acili ies, o example. On he o he hand, as he me hodology
is s anda dized, i is ans e able and applicable o o he ci ies. Mo eo e , his s udy lays
he g oundwo k o mo e ad anced me hods ha can u he explo e he complexi ies o
u ban phenomena. Fu u e esea ch could aim o c ea e mo e ep esen a i e samples o
he popula ion, educing biases in social ne wo k da a ela ed o age, gende , o o igin.
Ex ending he s udy pe iod o include seasonal a ia ions o mo e days o he week would
also allow o compa a i e s udies o occupancy pa e ns, p o iding a mo e accu a e
ep esen a ion o eali y and se ing as a aluable ool o ci y design and managemen .
Au ho Con ibu ions: Concep ualiza ion: Mikel Ba ena-He án, Ola z G ijalba and I zia Mod ego-
Mon o e; Da a cu a ion: Mikel Ba ena-He án; Fo mal analysis: Mikel Ba ena-He án; Funding
acquisi ion: Ola z G ijalba; In es iga ion: Mikel Ba ena-He án, Ola z G ijalba and I zia Mod ego-
Mon o e; Me hodology: Mikel Ba ena-He án; P ojec adminis a ion: Ola z G ijalba; So wa e:
Mikel Ba ena-He án; Resou ces: Mikel Ba ena-He án and I zia Mod ego-Mon o e; Supe ision:
Ola z G ijalba; Valida ion: Mikel Ba ena-He án, Ola z G ijalba and I zia Mod ego-Mon o e;
Visualiza ion: Mikel Ba ena-He án and I zia Mod ego-Mon o e; W i ing—o iginal d a : Mikel
Ba ena-He án, Ola z G ijalba and I zia Mod ego-Mon o e; W i ing— e iew and edi ing: Mikel
Ba ena-He án, Ola z G ijalba and I zia Mod ego-Mon o e. All au ho s ha e ead and ag eed o
he published e sion o he manusc ip .
ISPRS In . J. Geo-In . 2025,14, 221 21 o 23
Funding: This wo k was suppo ed by he Dipu ación Fo al Gipuzkoa unde g an numbe 2021-
CIEN-000044-05-02-01.
Da a A ailabili y S a emen : Publicly a ailable Google Popula Times da a, which uses agg ega ed
and anonymized da a om i s use s, was sc aped om he in e ne h ough he pla o m decla ed in
his pape . Me hodology o he ob en ion o he da a is also p o ided and, al hough no pos ed in a
eposi o y, ou da a can be accessed upon easonable eques .
Acknowledgmen s: This pape is pa o he p ojec “DinU : Mé odo de análisis de las dinámicas
u banas a a és de Big (Geo) Da a pa a la Regenación y T ans o mación de la ciudad” subsidized
by he Economic P omo ion and S a egic P ojec s o he Dipu ación Fo al Gipuzkoa h ough he
Gipuzkoa Science, Technology and Inno a ion Ne wo k P og amme.
Con lic s o In e es : The au ho s decla e no con lic s o in e es .
Re e ences
1. Häge s and, T. Wha abou people in Regional Science? Pap. Reg. Sci. Assoc. 1970,24, 6–21. [C ossRe ]
2.
Couclelis, H. Space, ime, geog aphy. In Geog aphical In o ma ion Sys ems, 1s ed.; Longley, P.A., Goodchild, M.F., Magui e, D.J.,
Rhind, D.W., Eds.; John Wiley and Sons: New Yo k, NY, USA, 1999; Volume 1, pp. 29–38.
3. Rapopo , A. Human Aspec s o U ban Fo m; Pe gamon P ess: Ox o d, UK, 1977. [C ossRe ]
4.
Hillie , B.; Penn, A.; Hanson, J.; G ajewski, T.; Xu, J. Na u al mo emen : O , con igu a ion and a ac ion in u ban pedes ian
mo emen . En i on. Plan. B Plan. Des. 1993,20, 29–66. [C ossRe ]
5.
Moo e-Che y, N.; Bonnin, C. Playing wi h ime in Moo e S ee , Dublin: U ban ede elopmen , empo al poli ics and he
go e nance o space- ime. U ban Geog . 2020,41, 1198–1217. [C ossRe ]
6.
Goodchild, M.F.; Janelle, D.G. The ci y a ound he clock: Space- ime pa e ns o u ban ecological s uc u e (Hali ax, Canada).
En i on. Plan. A 1984,16, 807–820. [C ossRe ]
7.
Janelle, D.G.; Klinkenbe g, B.; Goodchild, M.F. The empo al o de ing o u ban space and daily ac i i y pa e ns o popula ion
ole g oups. Geog . Sys . 1998,5, 117–137.
8.
Ande son, J. Space-Time Budge s and Ac i i y S udies in U ban Geog aphy and Planning. En i on. Plan. A Econ. Space 1971,
3, 353–368. [C ossRe ]
9.
Goodchild, M.F.; Klinkenbe g, B.; Janelle, D.G. A Fac o ial Model o Agg ega e Spa io-Tempo al Beha io : Applica ion o he
Diu nal Cycle. Geog . Anal. 1993,25, 277–294. [C ossRe ]
10. B ail, R.K.; Chapin, F.S. Ac i i y pa e ns o u ban esiden s. En i on. Beha . 1973,5, 163–190. [C ossRe ]
11. G ese, R.E. A Sense o Place, a Sense o Time. Landsc. J. 1995,14, 226–227. [C ossRe ]
12. Meen emeye , V. Geog aphical pe spec i es o space, ime, and scale. Landsc. Ecol. 1989,3, 163–173. [C ossRe ]
13.
Lísko ec, R.; Lich e , M.; Mulíˇcek, O. Ch ono opes o u ban cen ali ies: Looking o p ominen u ban imes and places. Geog . J.
2022,188, 166–176. [C ossRe ]
14.
Ne ejan, C.; Cunningham, S.; Se ka li, P. Ci y Rhy hm Logbook o an Explo a ion; Del Uni e si y o Technology: Del , The
Ne he lands, 2018.
15.
Chen, L.; Zhao, L.; Xiao, Y.; Lu, Y. In es iga ing he spa io empo al pa e n be ween he buil en i onmen and u ban ib ancy
using big da a in Shenzhen, China. Compu . En i on. U ban Sys . 2022,95, 101827. [C ossRe ]
16.
Wunde lich, F.M. Place-Tempo ali y and U ban Place-Rhy hms in U ban Analysis and Design: An Aes he ic Akin o Music.
J. U ban Des. 2013,18, 383–408. [C ossRe ]
17. Row, A.T.; Jacobs, J. The Dea h and Li e o G ea Ame ican Ci ies. Yale Law J. 1962,71, 1597–1602. [C ossRe ]
18.
Pa ka, E. Places as In e sec ing Flows: Mapping U ban Mo phologies, Func ional Cons ella ions and Pedes ian Rhy hms. In
P oceedings o he Space and Place, 4 h Global Con e ence, Ox o d, UK, 9–12 Sep embe 2013.
19.
Mulíˇcek, O.; Osman, R.; Seidenglanz, D. U ban hy hms: A ch ono opic app oach o u ban imespace. Time Soc. 2015,24, 304–325.
[C ossRe ]
20. Osman, R.; Mulíˇcek, O. U ban ch onopolis: Ensemble o hy hmized disloca ed places. Geo o um 2017,85, 46–57. [C ossRe ]
21.
Tan, W.; Klaasen, I. 24/7 En i onmen s: A Theo e ical and Empi ical Explo a ion om an U ban Planne s Pe spec i e. In
P oceedings o he Eu opean U ban Resea ch Associa ion (EURA): 10 h Anni e sa y Con e ence, Glasglow, Sco land, 12–14
Sep embe 2007.
22.
B omley, R.D.F.; Tallon, A.R.; Thomas, C.J. Disagg ega ing he space- ime laye s o ci y-cen e ac i i ies and hei use s. En i on.
Plan. A 2003,35, 1381–1851. [C ossRe ]
23.
Palme , J.R.B. Ac i i y-Space Seg ega ion: Unde s anding Social Di isions in Space and Time; P ince on Uni e si y: P ince on, NJ,
USA, 2013.

ISPRS In . J. Geo-In . 2025,14, 221 22 o 23
24.
Robe s, M.; Eld idge, A. Quie e , sa e , cheape : Planning o a mo e inclusi e e ening and nigh - ime economy. Plan. P ac . Res.
2007,22, 253–266. [C ossRe ]
25.
Van Liemp , I.; an Aals , I.; Schwanen, T. In oduc ion: Geog aphies o he u ban nigh . U ban S ud. 2015,52, 407–421. [C ossRe ]
26.
Robe s, M.; Tu ne , C. Con lic s o li eabili y in he 24-hou ci y: Lea ning om 48 hou s in he li e o London’s Soho. J. U ban
Des. 2005,10, 171–193. [C ossRe ]
27.
Gha ou i, A.; Webe , C. Mul i unc ional U ban Spaces a Solu ion o Inc ease he Quali y o U ban Li e in Dense Ci ies. Manza
2020,12, 34–45. [C ossRe ]
28.
Jones, M.; Pebley, A.R. Rede ining Neighbo hoods Using Common Des ina ions: Social Cha ac e is ics o Ac i i y Spaces and
Home Census T ac s Compa ed. Demog aphy 2014,51, 727–752. [C ossRe ]
29.
Meh a, V. Li ely s ee s: De e mining en i onmen al cha ac e is ics o suppo social beha io . J. Plan. Educ. Res. 2007,27, 165–187.
[C ossRe ]
30. Taylo , R.B. De ining Neighbo hoods in Space and Time. Ci yscape 2012,14, 225–230.
31.
Shi azi, M.R. Mapping neighbou hood ou doo ac i i ies: Space, ime, gende and age. J. U ban Des. 2019,24, 715–737. [C ossRe ]
32.
Smi h, R.J.; He he ing on, K. U ban hy hms: Mobili ies, Space and In e ac ion in he con empo a y ci y. Sociol. Re . 2013,61
(Suppl. 1), 4–16. [C ossRe ]
33.
Diepen, A.M.L.; Mus e d, S. Li es yles and he ci y: Connec ing daily li e o u bani y. J. Hous. Buil En i on. 2009,24, 331–345.
[C ossRe ]
34. Raman, S. Designing a li eable compac ci y physical o ms o ci y and social li e in u ban neighbou hoods. Buil En i on. 2010,
36, 63–80. [C ossRe ]
35. Boeing, G. Measu ing he complexi y o u ban o m and design. U ban Des. In . 2018,23, 281–292. [C ossRe ]
36.
Ba y, M.; Besussi, E.; Maa , K.; Ha s, J.J. Rep esen ing mul i unc ional ci ies: Densi y and di e si y in space and ime. Buil
En i on. 2004,30, 324–337. [C ossRe ]
37. Kwan, M.P. Time, in o ma ion echnologies, and he geog aphies o e e yday li e. U ban Geog . 2002,23, 471–482. [C ossRe ]
38.
Li, S.; D agice ic, S.; Cas o, F.A.; Ses e , M.; Win e , S.; Col ekin, A.; Pe i , C.; Jiang, B.; Hawo h, J.; S ein, A.; e al. Geospa ial big da a
handling heo y and me hods: A e iew and esea ch challenges. ISPRS J. Pho og amm. Remo e Sens. 2016,115, 119–133. [C ossRe ]
39.
Toole, J.L.; Colak, S.; S u , B.; Alexande , L.P.; E suko , A.; González, M.C. The pa h mos a eled: T a el demand es ima ion
using big da a esou ces. T ansp. Res. Pa C Eme g. Technol. 2015,58, 162–177. [C ossRe ]
40. Ki chin, R. The eal- ime ci y? Big da a and sma u banism. GeoJou nal 2014,79, 1–14. [C ossRe ]
41. Puebla, J.G. Big Da a y nue as geog a ías: La huella digi al de las ac i idades humanas. Doc. Anal. Geog . 2018,64, 195–217.
42.
Gao, S.; Liu, Y.; Wang, Y.; Ma, X. Disco e ing spa ial in e ac ion communi ies om mobile phone da a. T ans. GIS 2013,17, 463–481.
[C ossRe ]
43.
Liu, Z.; Ma, T.; Du, Y.; Pei, T.; Yi, J.; Peng, H. Mapping hou ly dynamics o u ban popula ion using ajec o ies econs uc ed om
mobile phone eco ds. T ans. GIS 2018,22, 494–513. [C ossRe ]
44.
Yang, X.; Fang, Z.; Xu, Y.; Shaw, S.-L.; Zhao, Z.; Yin, L.; Zhang, T.; Lin, Y. Unde s anding spa io empo al pa e ns o human
con e gence and di e gence using mobile phone loca ion da a. ISPRS In . J. Geoin . 2016,5, 177. [C ossRe ]
45. Gao, S. Spa io-Tempo al Analy ics o Explo ing Human Mobili y Pa e ns and U ban Dynamics in he Mobile Age. Spa . Cogn.
Compu . 2015,15, 86–114. [C ossRe ]
46.
Chen, Y.; Liu, X.; Li, X.; Liu, X.; Yao, Y.; Xu, X.; Pei, F. Delinea ing u ban unc ional a eas wi h building-le el social media da a: A
dynamic ime wa ping (DTW) dis ance based k-medoids me hod. Landsc U ban Plan. 2017,160, 48–60. [C ossRe ]
47.
Hu, Y.; Gao, S.; Janowicz, K.; Yu, B.; Li, W.; P asad, S. Ex ac ing and unde s anding u ban a eas o in e es using geo agged
pho os. Compu . En i on. U ban Sys . 2015,54, 240–254. [C ossRe ]
48.
Wang, Y.; Wang, T.; Tsou, M.H.; Li, H.; Jiang, W.; Guo, F. Mapping dynamic u ban land use pa e ns wi h c owdsou ced
geo- agged social media (Sina-Weibo) and comme cial poin s o in e es collec ions in Beijing, China. Sus ainabili y 2016,8, 1202.
[C ossRe ]
49.
S eige , E.; Resch, B.; Zip , A. Explo a ion o spa io empo al and seman ic clus e s o Twi e da a using unsupe ised neu al
ne wo ks. In . J. Geog . In . Sci. 2016,30, 1694–1716. [C ossRe ]
50.
Ma, X.; Liu, C.; Wen, H.; Wang, Y.; Wu, Y.J. Unde s anding commu ing pa e ns using ansi sma ca d da a. J. T ansp. Geog .
2017,58, 135–145. [C ossRe ]
51.
Gan, Z.; Yang, M.; Feng, T.; Timme mans, H. Unde s anding u ban mobili y pa e ns om a spa io empo al pe spec i e: Daily
ide ship p o iles o me o s a ions. T anspo a ion 2020,47, 315–336. [C ossRe ]
52.
F oehlich, J.; Neumann, J.; Oli e , N. Sensing and p edic ing he pulse o he ci y h ough sha ed bicycling. In P oceedings o he
IJCAI In e na ional Join Con e ence on A i icial In elligence, Pasadena, CA, USA, 11–17 July 2009.
53.
Ma, X.; Cao, R.; Jin, Y. Spa io empo al clus e ing analysis o bicycle sha ing sys em wi h da a mining app oach. In o ma ion 2019,
10, 163. [C ossRe ]
ISPRS In . J. Geo-In . 2025,14, 221 23 o 23
54.
Zeng, P.; Wei, M.; Liu, X. In es iga ing he spa io empo al dynamics o u ban i ali y using bicycle-sha ing da a. Sus ainabili y
2020,12, 1714. [C ossRe ]
55.
Rempe, F.; Hube , G.; Bogenbe ge , K. Spa io-Tempo al Conges ion Pa e ns in U ban T a ic Ne wo ks. T ansp. Res. P ocedia 2016,
15, 513–524. [C ossRe ]
56.
Zhang, K.; Sun, D.J.; Shen, S.; Zhu, Y. Analyzing spa io empo al conges ion pa e n on u ban oads based on axi GPS da a.
J. T ansp. Land Use 2017,10. [C ossRe ]
57.
Mao, F.; Ji, M.; Liu, T. Mining spa io empo al pa e ns o u ban dwelle s om axi ajec o y da a. F on . Ea h Sci. 2016,10,
205–221. [C ossRe ]
58. Ca pio-Pinedo, J.; Romanillos, G.; Apa icio, D.; Ma ín-Ca o, M.S.H.; Ga cía-Paloma es, J.C.; Gu ié ez, J. Towa ds a new u ban
geog aphy o expendi u e: Using bank ca d ansac ions da a o analyze mul i-sec o spa io empo al dis ibu ions. Ci ies 2022,
131, 103894. [C ossRe ]
59. Liu, S.; Zhang, L.; Long, Y. U ban i ali y a ea iden i ica ion and pa e n analysis om he pe spec i e o ime and space usion.
Sus ainabili y 2019,11, 4032. [C ossRe ]
60.
Xia, Z.; Li, H.; Chen, Y.; Liao, W. Iden i y and delimi a e u ban ho spo a eas using a ne wo k-based spa io empo al ield
clus e ing me hod. ISPRS In . J. Geoin . 2019,8, 344. [C ossRe ]
61.
Yao, X.; Zhu, D.; Gao, Y.; Wu, L.; Zhang, P.; Liu, Y. A S epwise Spa io-Tempo al Flow Clus e ing Me hod o Disco e ing Mobili y T ends;
IEEE Access: Pisca away, NJ, USA, 2018; Volume 6. [C ossRe ]
62.
San iago-Iglesias, E.; Ca pio-Pinedo, J.; Sun, W.; Ga cía-Paloma es, J.C. F ozen ci y: Analysing he dis up ion and esilience o
u ban ac i i ies du ing a hea y snow all e en using Google Popula Times. U ban Clim. 2023,51, 101644. [C ossRe ]
63. Cala io e, A.; Palme , G.; Combe , S.; A ibas-Bel, D.; Single on, A. A geog aphic da a science amewo k o he unc ional and
con ex ual analysis o human dynamics wi hin global ci ies. Compu . En i on. U ban Sys . 2021,85, 101539. [C ossRe ]
64.
Eus a . Tablas Es adís icas: Edi icios de la C.A. de Euskadi po Ámbi os Te i o iales, Según Tipo de Edi icio. Ins i u o Vasco
de Es adís ica. 2021. A ailable online: h ps://www.eus a .eus/municipal/da os_es adis icos/donos ia_san_sebas ian_c.h ml
(accessed on 30 July 2024).
65.
Fe nandez-C ehue , J.M.; Gimenez-Nadal, J.I.; Reyes Recio, L.E. The Na ional Wo k–Li e Balance Index©: The Eu opean Case.
Soc. Indic. Res. 2016,128, 341–359. [C ossRe ]
66.
Eus a . P opo ción de Pe sonas En e 16 y 74 Años Que Han Usado el Mó il en los Úl imos T es Meses. Published online 2022.
A ailable online: h ps://www.eus a .eus/indicado es/ emaseleccionado_5.b.1.1/ods2.h ml (accessed on 30 July 2024).
67.
Eus a . Pano ama de la Sociedad de la In o mación 2022 [No a de P ensa]. Ins i u o Vasco de Es adís ica. 2022. A ailable
online: h ps://www.eus a .eus/elemen os/el-565-de-los-in e nau as-pa icipa-en- edes-sociales-y-el-552-de-las-emp esas-
las-usa-pa a- ines-emp esa iales-en-la-ca-de-euskadi-en-2022/no 0020532_c.h ml (accessed on 30 July 2024).
68.
Eus a . Es ablecimien os y Pe sonas Empleadas en Municipios de Más de 10.000 Habi an es Según Rama de Ac i idad (A10).
Ins i u o Vasco de Es adís ica. 2023. A ailable online: h ps://www.eus a .eus/elemen os/ele0005800/es ablecimien os-y-pe sonas-
empleadas-en-municipios-de-mas-de-10000-habi an es-segun- ama-de-ac i idad-a10/ bl0005834_c.h ml (accessed on 30 July 2024).
69.
D’Zmu a, M. Behind he Scenes: Popula Times and Li e Busyness In o ma ion. A ailable online: h ps://blog.google/p oduc s/
maps/maps101-popula - imes-and-li e-busyness-in o ma ion/ (accessed on 3 Ma ch 2025).
70.
D obník, J. Google Maps Sc ape . 2019. A ailable online: h ps://gi hub.com/josiahakinloye/s o e-c awle -google-places
(accessed on 4 May 2022).
71.
Dipu ación de Gipuzkoa. Ca as o de Gipuzkoa. Dipu ación de Gipuzkoa. 2025. A ailable online: h ps://www.gipuzkoai ekia.
eus/es/da u-i ekien-ka alogoa/-/openDa aSea che /de ail/de ailView/ 249 d69-9765-4850-9b 8-c3ad457d848e (accessed on
24 Janua y 2025).
72. Dunham, H.W. The Image o he Ci y Ke in Lynch. Soc. P obl. 1960,8, 280–281. [C ossRe ]
73. Papa izos, J.; G a ano, L. K-Shape: E icien and Accu a e Clus e ing o Time Se ies. SIGMOD Rec. 2016,45, 69–76. [C ossRe ]
74.
Zhu, Y.; Diao, M. Unde s anding he spa io empo al pa e ns o public bicycle usage: A case s udy o Hangzhou, China. In . J.
Sus ain. T ansp. 2020,14, 163–176. [C ossRe ]
75.
San iago-Iglesias, E.; Romanillos, G.; Ca pio-Pinedo, J.; Sun, W.; Ga cía-Paloma es, J.C. Reco e ing u ban nigh li e: COVID-19
insigh s om Google Places ac i i y ends in Mad id. J. Maps 2024,20, 2371927. [C ossRe ]
76.
Happle, G.; Fonseca, J.A.; Schlue e , A. Con ex -speci ic u ban occupancy modeling using loca ion-based se ices da a. Build
En i on. 2020,175, 106803. [C ossRe ]
77.
Be nabeu-Bau is a, Á.; Se ano-Es ada, L.; Pe ez-Sanchez, V.R.; Ma í, P. The geog aphy o social media da a in u ban a eas:
Rep esen a i eness and complemen a i y. ISPRS In . J. Geoin . 2021,10, 747. [C ossRe ]
Disclaime /Publishe ’s No e: The s a emen s, opinions and da a con ained in all publica ions a e solely hose o he indi idual
au ho (s) and con ibu o (s) and no o MDPI and/o he edi o (s). MDPI and/o he edi o (s) disclaim esponsibili y o any inju y o
people o p ope y esul ing om any ideas, me hods, ins uc ions o p oduc s e e ed o in he con en .