mobile phone data analytics against the covid ...daily province-to-province ones into a relational...
TRANSCRIPT
CNR Fosca Giannotti Mirco Nanni Luca Pappalardo Giulio Rossetti Salvatore Rinzivillo
UNIVERSITY OF PISA Paolo Cintia Daniele Fadda Pietro Luigi Lopalco Sara Mazzilli Dino Pedreschi Lara Tavoschi
WINDTRE Pietro Bonato Francesco Fabbri Francesco Penone Marcello Savarese
MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALYFlow diversity and local job markets during the national lockdown
ISSUE #1April 2020
National Research Council of Italy
CNR, University of Pisa, WINDTRE
Page 2
TABLE OF CONTENTS
Mobile Phone Data
Origin-Destination matrices
Incoming, Outcoming and Internal Mobility Flows
Flow Diversity
Clusters of Provinces
Local Job Markets
Conclusion
References
Appendix
13
14
16
17
18
4
5
8
10
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 3
INTRODUCTION
—
Understanding human mobility patterns is crucial to plan the restart of production and eco-nomic activities, which are currently put in “stand-by” to fight the diffusion of the epidem-ics. A recent analysis shows that, following the national lockdown of March 9th, the mo-bility fluxes have decreased by 50% or more, everywhere in the country [13]. In this report, To this purpose, we use mobile phone data to compute the movements of people between Italian provinces, and we analyze the incoming, outcoming and internal mobility flows be-fore and during the national lockdown (March 9th, 2020) and after the closure of non-nec-essary productive and economic activities (March 23th, 2020). The population flow across provinces and municipalities enable for the modeling of a risk index tailored for the mobili-ty of each municipality or province. Such an index would be a useful indicator to drive coun-ter-measures in reaction to a sudden reactivation of the epidemics.
Mobile phone data, even when aggregated to preserve the privacy of individuals, are a use-ful data source to track the evolution in time of human mobility [8, 9], hence allowing for monitoring the effectiveness of control measures such as physical distancing [4, 5, 6]. In this report, we address the following analytical questions: How does the mobility structure of a territory change? Do incoming and outcoming flows become more predictable during the lockdown, and what are the differences between weekdays and weekends? Can we de-tect proper local job markets based on human mobility flows, to eventually shape the bor-ders of a local outbreak?
An interactive version of this report will be available at http://sobigdata.eu/covid_report
CNR, University of Pisa, WINDTRE
Page 4
MOBILE PHONE DATA
—
The raw data used in this report are the result of normal service operations performed by the mobile operator WINDTRE1: CDRs (Call Detail Records) and XDRs (eXtended Detail Records). In both cases, the fundamental geographical unit is the “phone cell” defined as the area covered by a single antenna, i.e., the device that captures mobile radio signals and keeps the user connected with the network. Multiple antennas are usually mounted on the same tower, each covering a different direction. The position of the tower (expressed as latitude and longitude) and the direction of the antenna allow inferring the extension of the corresponding phone cell. The position of caller and callee is approximat-ed by the corresponding antenna serving the call, whose extension is relatively small in urban contexts (in the order of 100m x 100m) and much larger in rural areas (in the order of 1km x 1km or more).
Based on this configuration, CDRs describe the location of mobile phone us-ers during call activities and XDRs their location during data transmission for internet access. The information content provided by standard CDR and XDR is the following:
In both CDRs and XDRs, the identity of the users is replaced by artificial iden-tifiers. The correspondence between such identifiers and the real identities of the users is known only to the mobile phone operator, who might use it in case of necessity. This pseudonymization procedure is a first important step (men-tioned in Article 6(4) and Article 25(1) of the GDPR, the EU General Data Protec-tion Regulation) to provide anonymity [7, 10, 11] and it will then turn into total-ly anonymous data for the possible treatment data use. For the analyses in this report, we used aggregated data computed by the mobile operator covering the period February 3rd, 2020 to March 28th, 2020.
For each phone call, a tuple <no , ni , t , As , Ae , d> is recorded, where no and ni are pseudo-anonymous identifiers, respectively of the “caller” and the “callee”; t is a timestamp saying when the call was placed; As and Ae are the identifiers of the towers/antennas to which the caller was connected at the start and end of the call; finally, d is the call duration (e.g., in minutes).
They are similar to CDRs, except that the communication is only between the antenna and the connected mobile phone, and an amount k of kilobytes is downloaded in the process. The format of XDR is, therefore, a tuple <n, t, A, k>.
Call Detail Records (CDR)
Extended Detail Records (XDR)
1WINDTRE is one of the main mobile phone operators in Italy, covering around 32% of the residential “human” mobile market.
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 5
ORIGIN-DESTINATIONMATRICES
—
CDRs and XDRs are aggregated into daily municipality-to-municipality origin-destination (OD) matrices: there is an OD matrix per each day, and each element ODA, B of the matrix describes the total number of trips from municipality A to municipality B. The presence of two consecutive points of a user in different municipalities indicates a movement, which is counted as a trip if the user stays in the destination municipality for at least one hour, and discarded otherwise. For a better matching with public COVID-19 data, we aggregated the municipality-to-municipality ODs into province-to-province ODs, in which each node repre-sents an Italian province. The trips between municipalities of the same province have been aggregated into a self-loop, which indicates the province’s internal mobility. As they are cal-culated by the operator, we store the daily municipality-to-municipality OD matrices and the daily province-to-province ones into a relational DBMS and access them through calls to a dedicated API.
Figure 1 visualizes the out-flows and in-flows of the province of Padua (region of Veneto, north-east of the country), for February 18th (before the lockdown, on the left) and March 24th (during the lockdown, on the right). The chart shows the flows among provinces with a stroke width proportional to the flows. The out-flows (in-flow) are first linked to the corre-sponding region and then to the final destination (origin). During the lockdown, we observe a drastic reduction of both the in- and the out-flows (reported on labels in the correspond-ing circle), as well as a reduction of the number of provinces the flows are coming from or are directed to. The reduction in the number of provenances and destinations is also evi-dent in the other provinces of the country. For example, Figure 2 shows this pattern is even more pronounced for the province of Bari, in the region of Puglia in the south-east of the country.
PADOVA2020-02-18
OR
IGIN
(0.015%) Bari
(0.015%) Bari
(0.028%) Napoli
(0.028%) Napoli
(0.010%) Pescara
(0.010%) Pescara
(0.203%) Rom
a
(0.203%) Rom
a(0.012%
) Viterbo
(0.012%) Viterbo
(0.012%) M
acerata
(0.012%) M
acerata
(0.038%) Ancona
(0.038%) Ancona
(0.041%) Pesaro,E,Urbino
(0.041%) Pesaro,E,Urbino
(0.021%) Terni
(0.021%) Terni
(0.032%) Perugia
(0.032%) Perugia
(0.022%) Prato
(0.022%) Prato
(0.025%) Siena
(0.025%) Siena
(0.030%) Arezzo
(0.030%) Arezzo
(0.039%) Pisa
(0.039%) Pisa
(0.022%) Livorno
(0.022%) Livorno
(0.168%) Firenze
(0.168%) Firenze
(0.024%) Pistoia
(0.024%) Pistoia
(0.033%) Lucca
(0.033%) Lucca
(0.107%) Rimini
(0.107%) Rimini
(0.124%) Forlì-cesena
(0.124%) Forlì-cesena
(0.173%) Ravenna
(0.173%) Ravenna
(0.881%) Ferrara
(0.881%) Ferrara
(1.063%) Bologna
(1.063%) Bologna
(0.251%) Modena
(0.251%) Modena
(0.113%) Reggio,Nell'emilia
(0.113%) Reggio,Nell'emilia
(0.086%) Parma(0.086%) Parma
(0.071%) Piacenza(0.071%) Piacenza
(0.010%) La,Spezia(0.010%) La,Spezia
(0.019%) Genova(0.019%) Genova
(0.473%) Pordenone(0.473%) Pordenone(0.217%) Trieste(0.217%) Trieste(0.117%) Gorizia(0.117%) Gorizia(0.526%) Udine(0.526%) Udine(9.328%) Rovigo(9.328%) Rovigo
(38.425%) Venezia
(38.425%) Venezia(13.647%) Treviso
(13.647%) Treviso(0.463%) Belluno
(0.463%) Belluno(24.992%) V
icenza
(24.992%) Vice
nza
(5.23
5%) V
erona
(5.23
5%) V
erona
(0.43
7%) T
rento
(0.43
7%) T
rento
(0.17
8%) B
olzan
o/boz
en
(0.17
8%) B
olzan
o/boz
en
(0.09
8%) M
onza
,E,D
ella,B
rianz
a
(0.09
8%) M
onza
,E,D
ella,B
rianz
a
(0.02
4%) L
odi
(0.02
4%) L
odi
(0.01
8%) L
ecco
(0.01
8%) L
ecco
(0.2
41%
) Man
tova
(0.2
41%
) Man
tova
(0.0
67%
) Cre
mon
a
(0.0
67%
) Cre
mon
a
(0.0
62%
) Pav
ia
(0.0
62%
) Pav
ia
(0.5
61%
) Bre
scia
(0.5
61%
) Bre
scia
(0.2
15%
) Ber
gam
o
(0.2
15%
) Ber
gam
o
(0.7
12%
) Mila
no
(0.7
12%
) Mila
no
(0.0
39%
) Com
o(0
.039
%) C
omo
(0.0
70%
) Var
ese
(0.0
70%
) Var
ese
(0.0
57%
) Ale
ssan
dria
(0.0
57%
) Ale
ssan
dria
(0.0
21%
) Cun
eo(0
.021
%) C
uneo
(0.0
30%
) Nov
ara
(0.0
30%
) Nov
ara
(0.0
12%
) Ver
celli
(0.0
12%
) Ver
celli
(0.0
65%
) Tor
ino
(0.0
65%
) Tor
ino
PugliaPuglia
Campania
Campania
AbruzzoAbruzzo
LazioLazio
Marche
Marche
UmbriaUmbria
Toscana
Toscana
Emilia,Romagna
Emilia,Romagna
LiguriaLiguriaFriuli,Venezia,GiuliaFriuli,Venezia,Giulia
VenetoVeneto
Trenti
no,Alto
,Adige
Trenti
no,Alto
,Adige
Lom
bard
iaLo
mba
rdia
Piem
onte
Piem
onte
(100.000%) Padova(100.000%) Padova(100.000%) Padova
Piem
onte
Lom
bard
ia
Trenti
no,Alto
,Adige
Veneto
Friuli,Venezia,GiuliaLiguriaEmilia,Romagna
ToscanaUmbria
Marche
Lazio
Abruzzo
CampaniaPuglia
(0.0
65%
) Tor
ino
(0.0
12%
) Ver
celli
(0.0
30%
) Nov
ara
(0.0
21%
) Cun
eo(0
.057
%) A
less
andr
ia(0
.070
%) V
ares
e(0
.039
%) C
omo
(0.7
12%
) Mila
no(0
.215
%) B
erga
mo
(0.5
61%
) Bre
scia
(0.0
62%
) Pav
ia(0
.067
%) C
rem
ona
(0.2
41%
) Man
tova
(0.01
8%) L
ecco
(0.02
4%) L
odi
(0.09
8%) M
onza
,E,D
ella,B
rianz
a
(0.17
8%) B
olzan
o/boz
en
(0.43
7%) T
rento
(5.23
5%) V
erona
(24.992%) Vice
nza
(0.463%) Belluno
(13.647%) Treviso
(38.425%) Venezia
(9.328%) Rovigo
(0.526%) Udine(0.117%) Gorizia(0.217%) Trieste(0.473%) Pordenone(0.019%) Genova(0.010%) La,Spezia
(0.071%) Piacenza(0.086%) Parma(0.113%) Reggio,Nell'emilia
(0.251%) Modena(1.063%) Bologna
(0.881%) Ferrara(0.173%) Ravenna
(0.124%) Forlì-cesena
(0.107%) Rimini
(0.033%) Lucca
(0.024%) Pistoia
(0.168%) Firenze
(0.022%) Livorno
(0.039%) Pisa
(0.030%) Arezzo
(0.025%) Siena
(0.022%) Prato
(0.032%) Perugia
(0.021%) Terni
(0.041%) Pesaro,E,Urbino
(0.038%) Ancona
(0.012%) M
acerata
(0.012%) Viterbo
(0.203%) Rom
a
(0.010%) Pescara
(0.028%) Napoli
(0.015%) Bari
0.015%0.015%0.028%0.028%
0.010%0.010%
0.215%0.215%
0.091%0.091%
0.053%0.053%
0.361%0.361%
2.868%2.868%
0.029%0.029%
1.333%1.333%
92.091%92.091%
0.615
%0.6
15%2.
106%
2.10
6%0.18
5%0.
185%
100.000%100.000%100.000%
0.18
5%
2.10
6%
0.615
%
92.091%
1.333%
0.029%
2.868%
0.361%0.053%
0.091%
0.215%
0.010%
0.028%
0.015%
DES
TIN
ATIO
N
Torino (0.066%)
Torino (0.066%)
Novara (0.021%
)N
ovara (0.021%)
Cuneo (0.013%
)C
uneo (0.013%)
Asti (0.015%)
Asti (0.015%)
Alessandria (0.032%)
Alessandria (0.032%)
Varese (0.033%)
Varese (0.033%)
Como (0.033%
)Com
o (0.033%)
Milano (0.632%
)
Milano (0.632%
)
Bergamo (0.238%
)
Bergamo (0.238%
)
Brescia (0.651%)
Brescia (0.651%)
Pavia (0.048%)
Pavia (0.048%)
Cremona (0.081%
)
Cremona (0.081%
)
Mantova (0.251%
)
Mantova (0.251%
)
Lecco (0.025%)
Lecco (0.025%)
Lodi (0.030%)
Lodi (0.030%)
Monza,E,Della,Brianza (0.087%)
Monza,E,Della,Brianza (0.087%)
Bolzano/bozen (0.173%)
Bolzano/bozen (0.173%)
Trento (0.424%)
Trento (0.424%)
Verona (5.220%)
Verona (5.220%)
Vicenza (25.092%)
Vicenza (25.092%)Belluno (0.534%)
Belluno (0.534%)Treviso (13.783%)
Treviso (13.783%)Venezia (38.002%)
Venezia (38.002%)Rovigo (9.339%)Rovigo (9.339%)Udine (0.582%)Udine (0.582%)Gorizia (0.105%)Gorizia (0.105%)Trieste (0.209%)Trieste (0.209%)
Pordenone (0.517%)Pordenone (0.517%)
Genova (0.028%)Genova (0.028%)
Piacenza (0.090%)Piacenza (0.090%)
Parma (0.092%)Parma (0.092%)
Reggio,Nell'emilia (0.140%)
Reggio,Nell'emilia (0.140%)
Modena (0.238%)
Modena (0.238%)
Bologna (1.004%)
Bologna (1.004%)
Ferrara (0.889%)
Ferrara (0.889%)
Ravenna (0.174%)
Ravenna (0.174%)
Forlì-cesena (0.129%)
Forlì-cesena (0.129%)
Rimini (0.145%)
Rimini (0.145%)
Lucca (0.030%)
Lucca (0.030%)
Pistoia (0.029%)
Pistoia (0.029%)
Firenze (0.181%)
Firenze (0.181%)
Livorno (0.018%)
Livorno (0.018%)
Pisa (0
.026%)
Pisa (0
.026%)
Arezzo (0
.039%)
Arezzo (0
.039%)
Siena (
0.025
%)
Siena (
0.025
%)
Prato (
0.034
%)
Prato (
0.034
%)
Perugia
(0.02
6%)
Perugia
(0.02
6%)
Pesa
ro,E,
Urbino
(0.04
6%)
Pesa
ro,E,
Urbino
(0.04
6%)
Anco
na (0
.040%
)
Anco
na (0
.040%
)
Macer
ata (0
.018%
)
Macer
ata (0
.018%
)
Asco
li,Pice
no (0
.010%
)
Asco
li,Pice
no (0
.010%
)Ro
ma
(0.1
95%
)
Rom
a (0
.195
%)
Tera
mo
(0.0
12%
)
Tera
mo
(0.0
12%
)Ca
serta
(0.0
13%
)
Case
rta (0
.013
%)
Napo
li (0.
035%
)Na
poli (
0.03
5%)
Bari
(0.0
25%
)Ba
ri (0
.025
%)
Brin
disi
(0.0
10%
)Br
indi
si (0
.010
%)
Cat
ania
(0.0
22%
)C
atan
ia (0
.022
%)
Piemonte
Piemonte
Lombardia
Lombardia
Trentino,Alto,Adige
Trentino,Alto,Adige
VenetoVeneto
Friuli,Venezia,GiuliaFriuli,Venezia,GiuliaLiguriaLiguria
Emilia,Romagna
Emilia,Romagna
Toscana
Toscana
Umbria
Umbria
March
eMar
che
Lazio
Lazio
Abru
zzo
Abru
zzo
Cam
pani
aCa
mpa
nia
Pugl
iaPu
glia
Sici
liaSi
cilia
(100.000%) Padova(100.000%) Padova(100.000%) Padova
Sici
liaPu
glia
Cam
pani
a
Abru
zzo
LazioMar
che
UmbriaTosca
na
Emilia,RomagnaLiguriaFriuli,Venezia,Giulia
Veneto
Trentino,Alto,Adige
Lombardia
Piemonte
Cat
ania
(0.0
22%
)
Brin
disi
(0.0
10%
)Ba
ri (0
.025
%)
Napo
li (0.
035%
)
Case
rta (0
.013
%)
Tera
mo
(0.0
12%
)
Rom
a (0
.195
%)
Asco
li,Pice
no (0
.010%
)
Macer
ata (0
.018%
)
Anco
na (0
.040%
)
Pesa
ro,E,
Urbino
(0.04
6%)
Perugia
(0.02
6%)
Prato (
0.034
%)
Siena (
0.025
%)
Arezzo (0
.039%)
Pisa (0
.026%)
Livorno (0.018%)
Firenze (0.181%)
Pistoia (0.029%)
Lucca (0.030%)Rimini (0.145%)
Forlì-cesena (0.129%)
Ravenna (0.174%)Ferrara (0.889%)
Bologna (1.004%)Modena (0.238%)Reggio,Nell'emilia (0.140%)Parma (0.092%)Piacenza (0.090%)
Genova (0.028%)
Pordenone (0.517%)Trieste (0.209%)Gorizia (0.105%)
Udine (0.582%)
Rovigo (9.339%)
Venezia (38.002%)
Treviso (13.783%)
Belluno (0.534%)
Vicenza (25.092%)
Verona (5.220%)
Trento (0.424%)
Bolzano/bozen (0.173%)
Monza,E,Della,Brianza (0.087%)
Lodi (0.030%)
Lecco (0.025%)
Mantova (0.251%
)
Cremona (0.081%
)Pavia (0.048%
)Brescia (0.651%
)Bergam
o (0.238%)
Milano (0.632%
)Com
o (0.033%)
Varese (0.033%)
Alessandria (0.032%)
Asti (0.015%)
Cuneo (0.013%
)N
ovara (0.021%)
Torino (0.066%)
0.147%0.147%
2.109%2.109%0.596%
0.596%
91.970%91.970%
1.413%1.413%
0.028%0.028%
2.900%2.900%
0.382%0.382%
0.026
%0.0
26%
0.114
%0.1
14%
0.19
5%0.
195%
0.01
2%0.
012%
0.04
8%0.
048%
0.03
6%0.
036%
0.02
2%0.
022%
100.000%100.000%100.000%
0.02
2%
0.03
6%
0.04
8%
0.01
2%
0.19
5%0.114
%
0.026
%0.382%
2.900%
0.028%
1.413%
91.970%
0.596%
2.109%
0.147%
CNR, University of Pisa, WINDTRE
Page 6
BEFORE LOCK DOWNDATA
—
Visualization of the in-flows and the out-flows of the province of Padua (region of Veneto, north-east of the country), on Tuesday February 18th (before the lockdown, on the left) and Tuesday March 24th (during the lockdown, on the right). Note that most of the flows are contained within the Padua’s region (Veneto) and neighboring regions, that the number of distinct origins and desti-nations of flows decrease during the lockdown.
PADOVA2020-03-24
OR
IGIN
(0.030%) Pesaro,E,U
rbino(0.030%
) Pesaro,E,Urbino
(0.043%) Perugia
(0.043%) Perugia
(0.028%) Arezzo
(0.028%) Arezzo
(0.033%) Livorno
(0.033%) Livorno
(0.106%) Firenze
(0.106%) Firenze
(0.033%) Lucca
(0.033%) Lucca
(0.028%) Rimini
(0.028%) Rimini
(0.126%) Forlì-cesena
(0.126%) Forlì-cesena
(0.166%) Ravenna
(0.166%) Ravenna
(0.527%) Ferrara
(0.527%) Ferrara
(0.615%) Bologna
(0.615%) Bologna
(0.209%) Modena
(0.209%) Modena
(0.116%) Reggio,Nell'emilia
(0.116%) Reggio,Nell'emilia
(0.101%) Parma
(0.101%) Parma
(0.053%) Piacenza
(0.053%) Piacenza
(0.038%) La,Spezia
(0.038%) La,Spezia
(0.413%) Pordenone
(0.413%) Pordenone
(0.101%) Trieste
(0.101%) Trieste
(0.076%) Gorizia(0.076%) Gorizia
(0.373%) Udine(0.373%) Udine
(11.224%) Rovigo(11.224%) Rovigo
(39.796%) Venezia(39.796%) Venezia
(11.753%) Treviso
(11.753%) Treviso
(0.184%) Belluno
(0.184%) Belluno(27.408%) Vicenza
(27.408%) Vicenza(4.77
5%) V
erona
(4.77
5%) V
erona
(0.22
2%) T
rento
(0.22
2%) T
rento
(0.15
9%) B
olzan
o/boz
en
(0.15
9%) B
olzan
o/boz
en
(0.05
0%) M
onza
,E,D
ella,B
rianz
a
(0.05
0%) M
onza
,E,D
ella,B
rianz
a
(0.2
72%
) Man
tova
(0.2
72%
) Man
tova
(0.0
83%
) Cre
mon
a
(0.0
83%
) Cre
mon
a
(0.0
38%
) Pav
ia
(0.0
38%
) Pav
ia
(0.3
78%
) Bre
scia
(0.3
78%
) Bre
scia
(0.1
01%
) Ber
gam
o
(0.1
01%
) Ber
gam
o
(0.2
32%
) Mila
no(0
.232
%) M
ilano
(0.0
45%
) Ale
ssan
dria
(0.0
45%
) Ale
ssan
dria
(0.0
28%
) Ast
i(0
.028
%) A
sti
(0.0
38%
) Nov
ara
(0.0
38%
) Nov
ara
Marche
Marche
Umbria
Umbria
ToscanaToscana
Emilia,Romagna
Emilia,Romagna
LiguriaLiguria
Friuli,Venezia,Giulia
Friuli,Venezia,Giulia
VenetoVenetoTre
ntino
,Alto
,Adig
e
Trenti
no,A
lto,A
dige
Lom
bard
iaLo
mba
rdia
Piem
onte
Piem
onte
(100.000%) Padova(100.000%) Padova(100.000%) Padova
Piem
onte
Lom
bard
ia
Trenti
no,A
lto,A
dige
Veneto
Friuli,Venezia,GiuliaLiguria
Emilia,RomagnaToscana
Umbria
Marche
(0.0
38%
) Nov
ara
(0.0
28%
) Ast
i(0
.045
%) A
less
andr
ia(0
.232
%) M
ilano
(0.1
01%
) Ber
gam
o(0
.378
%) B
resc
ia(0
.038
%) P
avia
(0.0
83%
) Cre
mon
a
(0.2
72%
) Man
tova
(0.05
0%) M
onza
,E,D
ella,B
rianz
a
(0.15
9%) B
olzan
o/boz
en
(0.22
2%) T
rento
(4.77
5%) V
erona
(27.408%) Vicenza
(0.184%) Belluno
(11.753%) Treviso
(39.796%) Venezia
(11.224%) Rovigo
(0.373%) Udine(0.076%) Gorizia(0.101%) Trieste(0.413%) Pordenone(0.038%) La,Spezia(0.053%) Piacenza
(0.101%) Parma
(0.116%) Reggio,Nell'emilia
(0.209%) Modena
(0.615%) Bologna
(0.527%) Ferrara
(0.166%) Ravenna
(0.126%) Forlì-cesena
(0.028%) Rimini
(0.033%) Lucca
(0.106%) Firenze
(0.033%) Livorno
(0.028%) Arezzo
(0.043%) Perugia
(0.030%) Pesaro,E,U
rbino0.030%0.030%
0.043%0.043%
0.199%0.199%
1.940%1.940%
0.038%0.038%
0.963%0.963%
95.141%95.141%
0.381
%0.3
81%1.
154%
1.15
4%0.11
1%0.
111%
100.000%100.000%100.000%
0.11
1%
1.15
4%
0.381
%
95.141%
0.963%0.038%
1.940%0.199%
0.043%
0.030%
DES
TIN
ATIO
N
Novara (0.030%
)N
ovara (0.030%)
Alessandria (0.052%)
Alessandria (0.052%)
Varese (0.035%)
Varese (0.035%)
Milano (0.200%
)M
ilano (0.200%)
Bergamo (0.139%
)
Bergamo (0.139%
)
Brescia (0.416%)
Brescia (0.416%)
Pavia (0.035%)
Pavia (0.035%)
Cremona (0.089%
)
Cremona (0.089%
)
Mantova (0.247%
)
Mantova (0.247%
)
Lodi (0.052%)
Lodi (0.052%)
Monza,E,Della,Brianza (0.062%)
Monza,E,Della,Brianza (0.062%)
Bolzano/bozen (0.109%)
Bolzano/bozen (0.109%)
Trento (0.233%)
Trento (0.233%)
Verona (4.903%)
Verona (4.903%)Vicenza (27.566%)
Vicenza (27.566%)
Belluno (0.200%)
Belluno (0.200%)
Treviso (11.766%)Treviso (11.766%)
Venezia (39.621%)Venezia (39.621%)
Rovigo (10.895%)Rovigo (10.895%)
Udine (0.423%)Udine (0.423%)
Gorizia (0.067%)
Gorizia (0.067%)
Trieste (0.141%)
Trieste (0.141%)
Pordenone (0.470%)
Pordenone (0.470%)
Genova (0.049%)
Genova (0.049%)
Piacenza (0.062%)
Piacenza (0.062%)
Parma (0
.089%)
Parma (0
.089%)
Reggio
,Nell'em
ilia (0
.084%
)
Reggio
,Nell'em
ilia (0
.084%
)
Moden
a (0.1
61%)
Moden
a (0.1
61%)
Bologn
a (0.5
86%)
Bologn
a (0.5
86%)
Ferra
ra (0.
510%
)
Ferra
ra (0.
510%
)
Rave
nna (
0.242
%)
Rave
nna (
0.242
%)
Forlì-
cese
na (0
.116%
)
Forlì-
cese
na (0
.116%
)Ri
mini
(0.0
27%
)
Rim
ini (0
.027
%)
Lucc
a (0
.040
%)
Lucc
a (0
.040
%)
Fire
nze
(0.1
09%
)
Fire
nze
(0.1
09%
)Li
vorn
o (0
.037
%)
Livo
rno
(0.0
37%
)Pi
sa (0
.030
%)
Pisa
(0.0
30%
)Ar
ezzo
(0.0
45%
)Ar
ezzo
(0.0
45%
)Si
ena
(0.0
32%
)Si
ena
(0.0
32%
)
Rom
a (0
.032
%)
Rom
a (0
.032
%)
Piemonte
Piemonte
Lombardia
Lombardia
Trentino,Alto,Adige
Trentino,Alto,AdigeVenetoVeneto
Friuli,Venezia,Giulia
Friuli,Venezia,Giulia
LiguriaLiguria
Emilia,R
omag
na
Emilia,R
omag
naTo
scan
aTo
scan
a
Lazi
oLa
zio
(100.000%) Padova(100.000%) Padova(100.000%) Padova
Lazi
o
Tosc
anaEmilia
,Rom
agna
LiguriaFriuli,Venezia,Giulia
Veneto
Trentino,Alto,Adige
Lombardia
Piemonte
Rom
a (0
.032
%)
Sien
a (0
.032
%)
Arez
zo (0
.045
%)
Pisa
(0.0
30%
)
Livo
rno
(0.0
37%
)
Fire
nze
(0.1
09%
)
Lucc
a (0
.040
%)
Rim
ini (0
.027
%)
Forlì-
cese
na (0
.116%
)
Rave
nna (
0.242
%)
Ferra
ra (0.
510%
)
Bologn
a (0.5
86%)
Moden
a (0.1
61%)
Reggio
,Nell'em
ilia (0
.084%
)
Parma (0
.089%)
Piacenza (0.062%)Genova (0.049%)
Pordenone (0.470%)Trieste (0.141%)
Gorizia (0.067%)Udine (0.423%)Rovigo (10.895%)
Venezia (39.621%)
Treviso (11.766%)
Belluno (0.200%)
Vicenza (27.566%)
Verona (4.903%)
Trento (0.233%)
Bolzano/bozen (0.109%)
Monza,E,Della,Brianza (0.062%)
Lodi (0.052%)
Mantova (0.247%
)
Cremona (0.089%
)Pavia (0.035%
)Brescia (0.416%
)Bergam
o (0.139%)
Milano (0.200%
)Varese (0.035%
)Alessandria (0.052%
)N
ovara (0.030%)
0.082%0.082%
1.274%1.274%
0.341%0.341%
94.951%94.951%
1.101%1.101%
0.049%0.049%
1.878
%1.8
78%
0.29
2%0.
292%
0.03
2%0.
032%
100.000%100.000%100.000%
0.03
2%
0.29
2%
1.878
%0.049%
1.101%
94.951%
0.341%
1.274%
0.082%
FIGURE 1 - PROVINCE OF PADUA
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 7
DURING LOCK DOWNDATA
—
BARI2020-02-18
OR
IGIN
(0.059%) C
agliari(0.059%
) Cagliari
(0.041%) C
rotone(0.041%
) Crotone
(0.059%) Reggio,Di,Calabria
(0.059%) Reggio,Di,Calabria
(0.057%) Catanzaro
(0.057%) Catanzaro
(0.430%) Cosenza
(0.430%) Cosenza
(7.866%) M
atera
(7.866%) M
atera(1.001%
) Potenza
(1.001%) Potenza
(39.423%) Barletta-andria-trani
(39.423%) Barletta-andria-trani
(3.672%) Lecce
(3.672%) Lecce
(17.748%) Brindisi
(17.748%) Brindisi
(18.979%) Taranto
(18.979%) Taranto
(4.475%) Foggia
(4.475%) Foggia
(0.303%) Salerno
(0.303%) Salerno
(0.247%) Avellino
(0.247%) Avellino
(0.718%) Napoli
(0.718%) Napoli
(0.088%) Benevento
(0.088%) Benevento
(0.283%) Caserta
(0.283%) Caserta
(0.231%) Campobasso
(0.231%) Campobasso
(0.238%) Chieti(0.238%) Chieti
(0.149%) Pescara(0.149%) Pescara
(0.127%) Teramo(0.127%) Teramo
(0.052%) L'aquila(0.052%) L'aquila
(0.072%) Frosinone(0.072%) Frosinone(0.034%) Latina(0.034%) Latina(1.229%) Roma(1.229%) Roma(0.025%) Fermo(0.025%) Fermo(0.059%) Ascoli,Piceno
(0.059%) Ascoli,Piceno(0.034%) Macerata
(0.034%) Macerata(0.125%) Ancona
(0.125%) Ancona(0.038%) Pesaro,E,Urbino
(0.038%) Pesaro,E,Urbino(0.032%) Perugia
(0.032%) Perugia(0.036%) Pisa
(0.036%) Pisa(0.086%) R
imini
(0.086%) Rimini
(0.057%) Forlì-c
esena
(0.057%) Forlì-c
esena
(0.190%) Bologna
(0.190%) Bologna
(0.05
2%) M
oden
a
(0.05
2%) M
oden
a
(0.02
9%) R
eggio
,Nell'em
ilia
(0.02
9%) R
eggio
,Nell'em
ilia
(0.04
5%) P
arma
(0.04
5%) P
arma
(0.04
3%) P
orden
one
(0.04
3%) P
orden
one
(0.06
1%) P
adov
a
(0.06
1%) P
adov
a
(0.10
9%) V
enez
ia
(0.10
9%) V
enez
ia
(0.0
95%
) Tre
viso
(0.0
95%
) Tre
viso
(0.0
41%
) Vice
nza
(0.0
41%
) Vice
nza
(0.0
43%
) Ver
ona
(0.0
43%
) Ver
ona
(0.0
41%
) Mon
za,E
,Del
la,B
rianz
a
(0.0
41%
) Mon
za,E
,Del
la,B
rianz
a
(0.0
27%
) Bre
scia
(0.0
27%
) Bre
scia
(0.1
09%
) Ber
gam
o
(0.1
09%
) Ber
gam
o
(0.6
77%
) Mila
no(0
.677
%) M
ilano
(0.0
34%
) Com
o(0
.034
%) C
omo
(0.0
88%
) Var
ese
(0.0
88%
) Var
ese
(0.0
63%
) Cun
eo(0
.063
%) C
uneo
(0.1
81%
) Tor
ino
(0.1
81%
) Tor
ino
SardegnaSardegnaCalabriaCalabria
BasilicataBasilicata
PugliaPuglia
CampaniaCampania
MoliseMolise
AbruzzoAbruzzo
LazioLazioMarcheMarche
UmbriaUmbriaToscana
Toscana
Emilia,Romagna
Emilia,Romagna
Friul
i,Ven
ezia,
Giulia
Friul
i,Ven
ezia,
Giulia
Vene
toVe
neto
Lom
bard
iaLo
mba
rdia
Piem
onte
Piem
onte
(100.000%) Bari(100.000%) Bari(100.000%) Bari
Piem
onte
Lom
bard
iaVe
neto
Friul
i,Ven
ezia,
Giulia
Emilia,Romagna
Toscana
Umbria
Marche
Lazio
AbruzzoMoliseCampania
Puglia
Basilicata
Calabria
Sardegna
(0.1
81%
) Tor
ino
(0.0
63%
) Cun
eo(0
.088
%) V
ares
e(0
.034
%) C
omo
(0.6
77%
) Mila
no(0
.109
%) B
erga
mo
(0.0
27%
) Bre
scia
(0.0
41%
) Mon
za,E
,Del
la,B
rianz
a
(0.0
43%
) Ver
ona
(0.0
41%
) Vice
nza
(0.0
95%
) Tre
viso
(0.10
9%) V
enez
ia
(0.06
1%) P
adov
a
(0.04
3%) P
orden
one
(0.04
5%) P
arma
(0.02
9%) R
eggio
,Nell'em
ilia
(0.05
2%) M
oden
a
(0.190%) Bologna
(0.057%) Forlì-c
esena
(0.086%) Rimini
(0.036%) Pisa
(0.032%) Perugia
(0.038%) Pesaro,E,Urbino
(0.125%) Ancona
(0.034%) Macerata
(0.059%) Ascoli,Piceno
(0.025%) Fermo
(1.229%) Roma
(0.034%) Latina(0.072%) Frosinone
(0.052%) L'aquila(0.127%) Teramo(0.149%) Pescara(0.238%) Chieti(0.231%) Campobasso(0.283%) Caserta
(0.088%) Benevento
(0.718%) Napoli(0.247%) Avellino
(0.303%) Salerno
(4.475%) Foggia(18.979%) Taranto
(17.748%) Brindisi
(3.672%) Lecce
(39.423%) Barletta-andria-trani
(1.001%) Potenza
(7.866%) M
atera
(0.430%) Cosenza
(0.057%) Catanzaro
(0.059%) Reggio,Di,Calabria
(0.041%) C
rotone
(0.059%) C
agliari0.059%0.059%
0.586%0.586%
8.867%8.867%
84.297%
84.297%1.639%1.639%
0.231%0.231%
0.566%0.566%
1.336%1.336%
0.281%0.281%0.032%0.032%
0.036%0.036%
0.460%0.460%0.0
43%
0.043
%0.34
9%0.
349%0.
976%
0.97
6%
0.24
4%0.
244%
100.000%100.000%100.000%
0.24
4%
0.97
6%
0.34
9%0.0
43%
0.460%
0.036%
0.032%
0.281%
1.336%
0.566%0.231%
1.639%
84.297%8.867%
0.586%
0.059%
DES
TIN
ATIO
N
Torino (0.104%)
Torino (0.104%)
Cuneo (0.050%
)C
uneo (0.050%)
Varese (0.117%)
Varese (0.117%)
Milano (0.332%
)M
ilano (0.332%)
Bergamo (0.115%
)
Bergamo (0.115%
)
Treviso (0.095%)
Treviso (0.095%)
Venezia (0.074%)
Venezia (0.074%)
Padova (0.036%)
Padova (0.036%)
Parma (0.034%
)
Parma (0.034%
)
Bologna (0.232%)
Bologna (0.232%)
Ravenna (0.025%)
Ravenna (0.025%)
Forlì-cesena (0.059%)
Forlì-cesena (0.059%)
Rimini (0.178%)
Rimini (0.178%)
Pisa (0.083%)
Pisa (0.083%)
Perugia (0.036%)
Perugia (0.036%)
Pesaro,E,Urbino (0.065%)
Pesaro,E,Urbino (0.065%)
Ancona (0.122%)
Ancona (0.122%)
Macerata (0.038%)
Macerata (0.038%)
Ascoli,Piceno (0.054%)
Ascoli,Piceno (0.054%)Fermo (0.025%)
Fermo (0.025%)Roma (1.002%)
Roma (1.002%)Frosinone (0.104%)
Frosinone (0.104%)L'aquila (0.047%)L'aquila (0.047%)Teramo (0.129%)Teramo (0.129%)
Pescara (0.210%)Pescara (0.210%)Chieti (0.280%)Chieti (0.280%)
Campobasso (0.232%)Campobasso (0.232%)
Isernia (0.027%)Isernia (0.027%)
Caserta (0.246%)Caserta (0.246%)
Benevento (0.079%)
Benevento (0.079%)
Napoli (0.702%)
Napoli (0.702%)
Avellino (0.336%)
Avellino (0.336%)
Salerno (0.284%)
Salerno (0.284%)
Foggia (4.767%)
Foggia (4.767%)
Taranto (19.437%)
Taranto (19.437%)
Brindis
i (17.7
02%)
Brindis
i (17.7
02%)
Lecc
e (3.7
00%)
Lecc
e (3.7
00%)
Barle
tta-a
ndria
-tran
i (38
.978
%)
Barle
tta-a
ndria
-tran
i (38
.978
%)
Pote
nza
(1.0
24%
)
Pote
nza
(1.0
24%
)M
ater
a (8
.178
%)
Mat
era
(8.1
78%
)Co
senz
a (0
.451
%)
Cose
nza
(0.4
51%
)Ca
tanz
aro
(0.0
56%
)
Cata
nzar
o (0
.056
%)
Regg
io,D
i,Cal
abria
(0.0
52%
)
Regg
io,D
i,Cal
abria
(0.0
52%
)Cr
oton
e (0
.029
%)
Crot
one
(0.0
29%
)
Cag
liari
(0.0
74%
)C
aglia
ri (0
.074
%)
Piemonte
Piemonte
Lombardia
Lombardia
VenetoVeneto
Emilia,Romagna
Emilia,Romagna
Toscana
ToscanaUmbriaUmbriaMarche
MarcheLazioLazio
AbruzzoAbruzzo
MoliseMolise
CampaniaCampania
Puglia
Puglia
Basil
icata
Basil
icata
Cala
bria
Cala
bria
Sard
egna
Sard
egna
(100.000%) Bari(100.000%) Bari(100.000%) Bari
Sard
egna
Cala
bria
Basil
icataPug
lia
CampaniaMolise
Abruzzo
Lazio
Marche
Umbria
ToscanaEmilia,Romagna
VenetoLom
bardiaPiem
onte
Cag
liari
(0.0
74%
)
Crot
one
(0.0
29%
)
Regg
io,D
i,Cal
abria
(0.0
52%
)
Cata
nzar
o (0
.056
%)
Cose
nza
(0.4
51%
)
Mat
era
(8.1
78%
)
Pote
nza
(1.0
24%
)
Barle
tta-a
ndria
-tran
i (38
.978
%)
Lecc
e (3.7
00%)
Brindis
i (17.7
02%)
Taranto (19.437%)Foggia (4.767%)Salerno (0.284%)
Avellino (0.336%)Napoli (0.702%)Benevento (0.079%)Caserta (0.246%)Isernia (0.027%)Campobasso (0.232%)
Chieti (0.280%)
Pescara (0.210%)
Teramo (0.129%)
L'aquila (0.047%)
Frosinone (0.104%)
Roma (1.002%)
Fermo (0.025%)
Ascoli,Piceno (0.054%)
Macerata (0.038%)
Ancona (0.122%)
Pesaro,E,Urbino (0.065%)
Perugia (0.036%)
Pisa (0.083%)
Rimini (0.178%)
Forlì-cesena (0.059%)
Ravenna (0.025%)
Bologna (0.232%)
Parma (0.034%
)Padova (0.036%
)Venezia (0.074%
)Treviso (0.095%
)Bergam
o (0.115%)
Milano (0.332%
)Varese (0.117%
)
Cuneo (0.050%
)Torino (0.104%
)0.153%0.153%
0.564%0.564%
0.205%0.205%0.528%
0.528%0.083%0.083%0.036%
0.036%0.305%0.305%
1.105%1.105%
0.665%0.665%
0.259%0.259%
1.647%1.647%
84.58
3%
84.58
3%
9.20
2%9.
202%
0.58
9%0.
589%
0.07
4%0.
074%
100.000%100.000%100.000%
0.07
4%
0.58
9%
9.20
2%84
.583%
1.647%
0.259%
0.665%
1.105%
0.305%
0.036%
0.083%0.528%
0.205%
0.564%
0.153%
CNR, University of Pisa, WINDTRE
Page 8
BEFORE LOCK DOWNDATA
—
Visualization of the in-flows and the out-flows of the province of Bari (in the region of Puglia, south-east of the country), February 18th (before the lockdown, on the left) and March 24th (dur-ing the lockdown, on the right). Note the drastic reduction in the number of distinct origins and destinations during the lockdown.
BARI2020-03-24
OR
IGIN
(0.376%) Cosenza
(0.376%) Cosenza
(7.553%) M
atera
(7.553%) M
atera(1.101%
) Potenza
(1.101%) Potenza
(38.299%) Barletta-andria-trani
(38.299%) Barletta-andria-trani
(1.967%) Lecce
(1.967%) Lecce
(23.612%) Brindisi
(23.612%) Brindisi
(20.054%) Taranto(20.054%) Taranto
(4.039%) Foggia(4.039%) Foggia
(0.446%) Salerno(0.446%) Salerno
(0.393%) Avellino
(0.393%) Avellino(0.402%) Napoli
(0.402%) Napoli(0.411%) Caserta
(0.411%) Caserta(0.254%) Campobasso
(0.254%) Campobasso(0.
358%
) Chie
ti
(0.35
8%) C
hieti
(0.18
4%) T
eramo
(0.18
4%) T
eramo
(0.2
19%
) Fro
sinon
e
(0.2
19%
) Fro
sinon
e
(0.2
36%
) Rom
a
(0.2
36%
) Rom
a
(0.0
96%
) Bol
ogna
(0.0
96%
) Bol
ogna
CalabriaCalabria
BasilicataBasilicata
PugliaPuglia
CampaniaCampaniaMoliseMoliseAbru
zzo
Abruzz
oLazio
Lazio
Emilia
,Rom
agna
Emilia
,Rom
agna
(100.000%) Bari(100.000%) Bari(100.000%) Bari
Emilia
,Rom
agna
Lazio
Abruzz
o
Molise
Campania
Puglia
Basilicata
Calabria
(0.0
96%
) Bol
ogna
(0.2
36%
) Rom
a(0
.219
%) F
rosin
one
(0.18
4%) T
eramo
(0.35
8%) C
hieti
(0.254%) Campobasso
(0.411%) Caserta
(0.402%) Napoli
(0.393%) Avellino
(0.446%) Salerno
(4.039%) Foggia
(20.054%) Taranto
(23.612%) Brindisi
(1.967%) Lecce(38.299%) Barletta-andria-trani
(1.101%) Potenza
(7.553%) M
atera
(0.376%) Cosenza
0.376%0.376%
8.655%8.655%
87.971%87.971%
1.652%1.652%
0.254%0.254%
0.542
%0.5
42%0.
455%
0.45
5%0.09
6%0.
096%
100.000%100.000%100.000%
0.09
6%
0.45
5%
0.542
%
0.254%
1.652%
87.971%
8.655%
0.376%
DES
TIN
ATIO
N
Forlì-cesena (0.094%)
Forlì-cesena (0.094%)
Pesaro,E,Urbino (0.112%)
Pesaro,E,Urbino (0.112%)
Ancona (0.103%)
Ancona (0.103%)
Roma (0.489%)
Roma (0.489%)
Frosinone (0.146%)
Frosinone (0.146%)
Teramo (0.257%)
Teramo (0.257%)
Chieti (0.438%)
Chieti (0.438%)Campobasso (0.326%)
Campobasso (0.326%)Caserta (0.257%)
Caserta (0.257%)Benevento (0.137%)
Benevento (0.137%)Napoli (0.386%)Napoli (0.386%)
Avellino (0.523%)Avellino (0.523%)
Salerno (0.403%)Salerno (0.403%)
Foggia (4.290%)Foggia (4.290%)
Taranto (19.964%)
Taranto (19.964%)
Brindisi (22.615%)
Brindisi (22.615%)
Lecce
(2.11
1%)
Lecce
(2.11
1%)
Barle
tta-a
ndria
-tran
i (37.7
06%
)
Barle
tta-a
ndria
-tran
i (37.7
06%
)Po
tenz
a (1
.098
%)
Pote
nza
(1.0
98%
)M
ater
a (8
.082
%)
Mat
era
(8.0
82%
)
Cose
nza
(0.4
63%
)Co
senz
a (0
.463
%)
Emilia,Rom
agnaEm
ilia,Romagna
Marche
Marche
LazioLazioAbruzzo
AbruzzoMoliseMolise
CampaniaCampania
PugliaPuglia
Basil
icata
Basil
icata
Cala
bria
Cala
bria
(100.000%) Bari(100.000%) Bari(100.000%) Bari
Cala
bria
Basil
icata
Puglia
Campania
Molise
Abruzzo
LazioM
arche
Emilia,Rom
agna
Cose
nza
(0.4
63%
)
Mat
era
(8.0
82%
)
Pote
nza
(1.0
98%
)
Barle
tta-a
ndria
-tran
i (37.7
06%
)
Lecce
(2.11
1%)
Brindisi (22.615%)
Taranto (19.964%)
Foggia (4.290%)
Salerno (0.403%)
Avellino (0.523%)
Napoli (0.386%)
Benevento (0.137%)
Caserta (0.257%)
Campobasso (0.326%)
Chieti (0.438%)
Teramo (0.257%)
Frosinone (0.146%)
Roma (0.489%)
Ancona (0.103%)
Pesaro,E,Urbino (0.112%)
Forlì-cesena (0.094%)
0.094%0.094%
0.214%0.214%0.635%
0.635%
0.695%0.695%
0.326%0.326%
1.707%1.707%
86.685%86.685%
9.18
0%9.
180%
0.46
3%0.
463%
100.000%100.000%100.000%
0.46
3%
9.18
0%
86.685%
1.707%
0.326%
0.695%
0.635%
0.214%
0.094%
FIGURE 2 - PROVINCE OF BARI
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 9
DURING LOCK DOWNDATA
—
The two vertical lines indicate the dates of the national lockdown (March 9th) and the closure of non-necessary productive and economic activities (March 23th). We observe a significant decrease in the volume of flows after the national lockdown, while we do not observe a comparable decrease soon after the closure of non-necessary activities.
FIGURE 3 - EVOLUTION OF FLOWS
CNR, University of Pisa, WINDTRE
Page 10
INCOMING, OUTCOMING AND INTERNAL MOBILITY FLOWS
—
How did mobility flows in Italy change during the lockdown? This aspect is crucial to quantify to what extent the government direc-tives have had the desired effect. We selected four Italian provinc-es: Bergamo, Padua, Bari, and Catania. For each of these provinces, we computed the evolution day by day of three types of flow:
Figure 3 shows the evolution of the normalized in-flows, out-flows, and self-flows of the selected provinces. It is evident how all prov-inces have a net decrease of the in-, out- and self-flows soon af-ter the first national lockdown (March 9th), and a stabilization of the flows on the new volume after around one week, from March 15th. Therefore, subsequent ordinances, such as closing factories on March 17th, have had a minor impact on the reduction of mo-bility flows.
1. out-flows, indicating the total number of people moving from the province to any other province in Italy on that day;
2. in-flows, indicating the total number of people moving to the province from any other province in Italy on that day;
3. self-flows (or internal flows), indicating the total number of people moving between municipalities of the same province on that day.
NORMALIZED FLOWS WITH RESPECT TO THE MAXIMUM VOLUME OBSERVED
CATANIA
BARI
PADUA
BERGAMO
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 11
FLOWDIVERSITY
—
An important aspect of the mobility of a province is the diversification of the provenience and the destination of people. Specifically, we define the in-flow diversity of a province A as the Shannon entropy of the in-flows to the province [12]:
where Pin is the number of provinces with non-null flow to province A, p(x) is the probabili-ty that the in-flow to province A comes from province x, and log(N) is a normalization fac-tor where N=110 is the number of Italian provinces. The out-flow diversity of province A is computed similarly as:
where Pout is the number of provinces with non-null flow from province A, and p(x) is the probability that the out-flow from province A goes to province x.
The horizon charts in Figure 4 show the evolution of the in- and out-flow diversity for the four selected provinces, while those in Figure 5 refer to 30 provinces chosen randomly. The vertical axis lines represents time, each rectangle section has a color proportional to the displayed measure (darker color for larger value). The circles on the left have an area pro-portional to the number of confirmed COVID-19 cases in the corresponding province up to March 24th. We find a progressive reduction of both the in- and out-flow diversity as time goes by, with an acceleration of the process soon after the beginning of the national lock-down (March 9th). Before the lockdown, the in- and out-flow diversities are slightly higher at the weekends than the weekdays. The opposite is true during the lockdown: the in- and out-flow diversities are considerably lower at the weekends than the weekdays. This ex-citing result suggests that: (i) the provenience and the destination of a province’s mobili-ty flows during the lockdown are more predictable than before the lockdown; (ii) regarding the weekends, the provenience and destination of flows are more diverse before the lock-down than during it.
CNR, University of Pisa, WINDTRE
Page 12
BERGAMO (num. casi 6728)
PADOVA (num. casi 1464)
BARI (num. casi 304)
CATANIA (num. casi 286)
Feb 09 Feb 16 Feb 23 March Mar 08 Mar 15 Mar 22INFLOWNational Lockdown
2020-03-08
Closure non necessary activities
2020-03-22
Series 0
Series 1
Series 2
Series 3
BERGAMO (num. casi 6728)
PADOVA (num. casi 1464)
BARI (num. casi 304)
CATANIA (num. casi 286)
Feb 09 Feb 16 Feb 23 March Mar 08 Mar 15 Mar 22OUTFLOWNational Lockdown
2020-03-08
Closure non necessary activities
2020-03-22
Series 0
Series 1
Series 2
Series 3
Horizon chart that describes the evolution in time of the in- and out-flow diversity of the provinces of Bergamo, Padua, Bari and Catania. The circles on the left have an area proportional to the number of confirmed COVID-19 cases in the corresponding province up to March 24th. Horizon charts compact the area chart by slicing it horizontally, and then shifting the slices to baseline zero. Black solid ver-tical lines indicate the dates of the national lockdown (March 9th) and the closure of non-necessary productive and economic activities (March 23th). The white dashed vertical lines indicate Sundays. Note that, while the in- and out-flow diversities slightly increase in the weekends before the lockdown, they decrease in the weekends during the lockdown.
FIGURE 4 - EVOLUTION OF FLOW DIVERSITY
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 13
Horizon chart that describes the evolution in time of the in-flow diversity of 30 provinces (out of 110) chosen at random. Solid vertical lines indicate the dates of the lockdown (March 9th) and the closure of non-necessary economic activities (March 23th). The dashed vertical lines indicate Sundays. We observe an inter-esting pattern for weekends: while flow diversity slightly increases with respect to weekdays before the lockdown, it decreases during the lockdown.
FIGURE 5 - EVOLUTION OF FLOW DIVERSITY
CNR, University of Pisa, WINDTRE
Page 14
Evolution of the in-flow diversity of provinces in cluster 4. This cluster includes Bergamo and Brescia, territories among the most hit by COV-ID-19. The circles on the left have an area pro-portional to the number of confirmed cases in the corresponding province up to March 24th.
Evolution of the in-flow diversity of the five clusters’ centroids. The area around the line indicates the deviation of the provinces from the centroid. Note that, though the clusters have similar trends, they have different typical in-flow diversities.
CLUSTERSOF PROVINCES
—
We use the k-means clustering algo-rithm to discover k groups of similar provinces in terms of their evolution of in- and out-flow diversity. To find the best value of k, we repeat the algorithm for k = 2, ..., 20. For both the in- and out-flow diversities, we find that k = 5 min-imizes the within-cluster distance. Fig-ure 6 shows the centroids of the five clusters of in-flow diversity.
Although the clusters’ trends are sim-ilar, they have different typical in-flow diversities. We provide in the Appen-dix the figure that shows the clusters’ centroids regarding the evolution of the out-flow diversity. Figure 7 visualiz-es the evolution of the in-flow diversi-ty for all the provinces in cluster 4, the one with the highest typical in-flow di-versity. We provide in the Appendix the horizon charts regarding the other four clusters.
FIGURE 6 - CLUSTERS OF PROVINCES
FIGURE 7 - EVOLUTION OF IN-FLOW DIVER-SITY OF CLUSTER 4
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 15
LOCAL JOB MARKETS
—
Economic activities are linked by input-output relationships, with interconnected supply chains that are difficult to isolate. Local Job Markets (LJM) take into account the shifts be-tween the home-work displacements (commuting) that occur between different municipal-ities. Each LJM is partially isolated from the others, allowing from a more precise control with respect to the administrative classifications of the territory (e.g., municipalities, prov-inces, regions).
One strategy to detect LJMs is to identify clusters of geographical areas that are homo-geneous w.r.t. their mobility flows, i.e., groups of municipalities for which the internal mo-bility fluxes are denser than the outgoing ones. This can be done by analyzing the munici-pality-to-municipality OD matrices as weighted directed graphs [1] and using a community detection algorithm [2] to discover a collection of well-bounded mesoscale topologies, e.g., municipality clusters. Note that community detection algorithms can provide different re-sults depending on their definition of what a community is [2]. We use algorithm Infomap [3], which uses the probability flow of random walks on a graph as a proxy for information flows in the real system, and decomposes the network into clusters by compressing a de-scription of the probability flow. The algorithm looks for a cluster partition of the given net-work that minimizes the expected description length of a random walk.
We applied Infomap to the daily municipality-to-municipality OD matrices. Figure 8 shows the evolution in time of the number of LJMs (communities) in the country. Since the na-tional lockdown (March 9th), there has been a striking increase in the number of communi-ties, indicating that people moved within smaller areas. For example, note that on Monday, March 2nd (before the lockdown), we have around 350 communities, while the number of communities on Monday, March 16th (during the lockdown) is around 550 communities, 200 more. We also find that, during the lockdown, the number of communities increases by around 50% on the weekends (in Figure 8, Sundays are denoted by vertical lines). Also, note that the number of communities increases after the closure of non-necessary productive and economic activities (March 22nd). This may indicate that, while we do not appreciate any significant difference in the volume and diversity of flows after this closure, the struc-ture of mobility flows has changed significantly.
CNR, University of Pisa, WINDTRE
Page 16
Figure 9 shows the local job markets we found in Puglia (a region on the south-east of the country) before the lockdown (up) and during the lockdown (down). Note the fragmentation of the territory during the lockdown, especially for the easternmost and the westernmost parts of the region.
Evolution in time of the number of Local Job Markets (communities) in Italy, according to the Infom-ap algorithm. Grey vertical lines indicate Sundays. Note that, after the beginning of the national lock-down (March 9th), there is a striking increase of the LJMs. Moreover, there is a slight increase of the communities after the closure of non-necessary activities (March 22nd).
Local job markets in Puglia (a region in the south-east of the country) before the lock-down (left) and during the lockdown (right).
FIGURE 8 - NUMBER OF LOCAL JOB MARKETS IN TIME
FIGURE 9 - LOCAL JOB MARKETS IN PUGLIA.
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 17
CONCLUSIONS
—
In our first report of the analysis of mobility flows using mobile phone data up to March 28th, 2020, we find several interesting results. First, regarding the volume of in-, out- and self-flows between provinces, we find a significant decrease after the national lockdown (March 9th). Still, we do not find any significant decrease soon after the closure of the non-necessary productive activities.
Regarding the in- and out-flow diversities, we find that while there is a slight increase in the flow diversity on the weekends before the lockdown, there is a strong decrease of the flow diversity on the weekends during the lockdown. Moreover, the application of data mining techniques reveals the presence of five main clusters of provinces.
Finally, we use a community detection algorithm to find local job markets in Italy. We ob-serve a striking increase in the number of communities during the lockdown and a slight increase after the closure of non-necessary activities. This suggests that reduced mobility split the territory into more and smaller local job markets. This information may be exploit-ed by decision- and policy-makers to plan “phase 2” of the management of the epidemics.
In the next report, we will investigate deeply how the structure of the OD matrices evolve in time, and we will extend the period of observation to the most recent days. We will also focus our analysis on some specific regions, considering the evolution of the epidemics at a municipality level. We will compare the impact of mobility reduction to the outbreak, answering several analytical questions: What is the virus-spreading effect generated by late-February north-south flows? How large should a “red zone” be to reduce effectively the spread of the epidemic?
CNR, University of Pisa, WINDTRE
Page 18
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 19
REFERENCES
—
[1] Rinzivillo, Salvatore, et al. “Discovering the geographical borders of human mobility.” KI-Künstliche Intelligenz 26.3 (2012): 253-260.
[2] Fortunato, Santo. “Community detection in graphs.” Physics reports 486.3-5 (2010): 75-174.
[3] Rosvall, Martin, and Carl T. Bergstrom. “An information-theoretic framework for resolving community structure in complex networks.” Proceedings of the National Academy of Sciences 104.18 (2007): 7327-7331.
[4] Oliver, Nuria, et al., “Mobile phone data and COVID-19: Missing an opportunity?” arXiv e-prints arXiv:2003.12347 (2020).
[5] Buckee, Caroline O., et al., “Aggregated mobility data could help fight COVID-19”, Science 368:6487 (2020), 145-146.
[6] Ferres, Leo, et al., “Measuring Levels of Activity in a Changing City: A Study Using Cellphone Data Streams”, http://datascience.udd.cl/covid_ids_tef_01.pdf
[7] de Montjoye, Y., et al. “On the privacy-conscientious use of mobile phone data”. Scientific Data 5: 180286 (2018).
[8] Blondel, V. D., et al., “A survey of results on mobile phone datasets analysis”, EPJ Data Science 4:10 (2015). https://doi.org/10.1140/epjds/s13688-015-0046-0
[9] Pappalardo, L., et al., “Returners and explorers dichotomy in human mobility”, Nature Communications 6:8166 (2015). https://doi.org/10.1038/ncomms9166
[10] Pratesi, F., et al., “PRIMULE: Privacy risk mitigation for user profiles”, Data & Knowledge Engineering 125: 101786 (2020). https://doi.org/10.1016/j.datak.2019.101786
[11] Pellungrini, R., et al., “A Data Mining Approach to Assess Privacy Risk in Human Mobility Data”, ACM Transactions in. Intelligent Systems and Technologies 9:3 (2017). https://doi.org/10.1145/3106774
[12] Pappalardo, L., et al., “An analytical framework to nowcast well-being using mobile phone data”, International Journal of Data Science Analytics 2, 75–92 (2016). https://doi.org/10.1007/s41060-016-0013-2 [13] Pepe et al., “COVID-19 outbreak response: first assessment of mobility changes in Italy following lockdown”, https://covid19mm.github.io/in-progress/2020/03/13/first-report-assessment.html (2020).
APPENDIX
—
Evolution of the out-flow diversity of the five clusters’ centroids. The area around the line indicates the deviation of the provinc-es from the centroid. Note that, though the clusters have similar trends, they have dif-ferent typical out-flow diversities.
Figure. Evolution of the in-flow diversity of provinces in cluster 1 This cluster includes Bergamo and Brescia, territories among the most hit by COVID-19. The circles on the left have an area proportional to the number of confirmed cases in the corresponding province up to March 24th.
FIGURE 10
FIGURE 11
CNR, University of Pisa, WINDTRE
Page 20
Evolution of the in-flow diversity of provinces in cluster 3. The circles on the left have an area propor-tional to the number of confirmed COVID-19 cases in the corresponding province up to March 24th.
FIGURE 12
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 21
Evolution of the in-flow diversity of provinces in cluster 0. The circles on the left have an area propor-tional to the number of confirmed COVID-19 cases in the corresponding province up to March 24th.
FIGURE 13
CNR, University of Pisa, WINDTRE
Page 22
Evolution of the in-flow diversity of provinces in cluster 2. The circles on the left have an area propor-tional to the number of confirmed COVID-19 cases in the corresponding province up to March 24th.
FIGURE 14
Flow diversity and local job markets during the national lockdown MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19 EPIDEMICS IN ITALY
Page 23
CNR
―ISTI
Fosca Giannotti
Mirco Nanni Luca Pappalardo
Giulio Rossetti Salvatore Rinzivillo
UNIVERSITY OF PISA
―COMPUTER SCIENCE
DEPA RTMENT
Paolo Cintia Daniele Fadda
Dino Pedreschi
―DEPA RTMENT
OF TR A NSL ATION A L RESE A RCH
ON NE W TECHNOLOGIES IN MEDICINE A ND SURGERY
Pier Luigi Lopalco
Sara Mazzilli Lara Tavoschi
WINDTRE
―BIG DATA
& ANALYTICS
Pietro Bonato Francesco Fabbri
Francesco Penone Marcello Savarese
MOBILE PHONE DATA ANALYTICS AGAINST THE COVID-19
EPIDEMICS IN ITALY
Flow diversity and local job markets during the national lockdown
ISSUE #1April 2020
National Research Council of Italy
for more information [email protected], [email protected],
[email protected], [email protected]
Released under Creative Commons Attribution-Noncommercial-ShareAlike license (CC BY-NC-SA 4.0)