Database Reference
In-Depth Information
Movements trend during the day
700
600
500
400
300
200
100
0
(a)
3500
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
3000
2500
2000
1500
1000
500
0
Frequency of
movenents
(b)
Figure 10.1 Movement distribution by hour: (a) representative weekday in MilanoSurvey
and (b) entire week in Milano2007.
we measure the number of moving vehicles in every hour of the day, and create
a histogram over the entire week. The result is shown in Figure 10.1 together
with a typical day distribution provided by the MilanoSurvey.
The two distributions match significantly, especially for the days from the
second to the fifth of the week, which actually represent regular working days,
fromMonday to Thursday. Background knowledge and domain expertise, in this
case, help to explain the anomaly of Friday. Indeed, it was Easter Friday, which
explains the different shape with respect to previous weekdays. Within working
days, the most relevant deviation from the survey data is a higher volume of
movement between the two peaks in the rush hours and (to a minor extent) the
later part of the day. The assessment with the mobility agency revealed that
the results are coherent, and actually in this specific case GSP data prove to
be more robust than surveys. Indeed, the latter are known to underestimate the
movements in these periods of the day, due to the fact that interviewed people
tend not to report their occasional mobility, such as going to the dentist or visiting
a friend. Also, GPS data contain mobile activity of people who do not live in the
greater metropolitan area, while the survey focuses on Milano residents.
The second dimension of analysis is the spatial component. Here we can
try to estimate the presence of population on the territory through GPS data,
and compare it with correspondent results obtained from the declared places of
residence on the surveys (see Figure 10.2 a). A similar estimate was obtained on
Milano2007 by (1) partitioning the space into a regular grid and (2) counting for
each cell the number of vehicles that were stationary in the cell for each time
Search WWH ::




Custom Search