Detecting the persistent MFAs

• Author: Sermi, F.; Iacus, S.; Vespe, M.; Santamaria, C.; Spyratos, S.; Tarchi, D.
• Pages: 7-10

Figure 2 is a heatmap representation of the matrix of similarity index among all MFAs for

the period 1 February 2020 - 10 June 2020. Darkest-bluish zones are most different,

whereas lighter-reddish are the most similar. It is interesting to observe the difference

between pre- and post- lockdown (14 March 2020), then a slow recovery to normality. It

is also worth noticing that holidays and weekends have clearly different mobility patterns

than weekdays and that these MFAs are differ from the administrative borders (Spanish

provinces), especially during weekdays (see Figure 3).

2020−02−01

2020−02−08

2020−02−15

2020−02−22

2020−02−29

2020−03−07

2020−03−14

2020−03−21

2020−03−28

2020−04−04

2020−04−11

2020−04−18

2020−04−25

2020−05−02

2020−05−09

2020−05−16

2020−05−23

2020−05−30

2020−06−06

2020−02−01

2020−02−08

2020−02−15

2020−02−22

2020−02−29

2020−03−07

2020−03−14

2020−03−21

2020−03−28

2020−04−04

2020−04−11

2020−04−18

2020−04−25

2020−05−02

2020−05−09

2020−05−16

2020−05−23

2020−05−30

2020−06−06

0.00

0.25

0.50

0.75

1.00

Similarity

Weekday

holiday

working

Similarity among all days

Figure 2: Similarity index matrix of all MFAs in Spain. Period: 1 February 2020 - 10 June 2020.

4 Detecting the persistent MFAs

As seen in the previous section, the MFAs have daily patterns, they change between before

and after the lockdown is in force and tend to go back to their original shapes after the

ease of containment measures. Moreover, MFAs shows time-variability also for the same

weekday; thus, in order to fully exploit their potential, a stable version of the MFAs needs

to be identified. Since the number of MFAs changes day by day and the same cell may

move from an MFA to another (changing the MFA label associated to it), we apply a CO-

association method. The C O-association method (CO ) avoids the label correspondence

problem. It does so by mapping the ensemble members onto a new representation where

7