2. MARCO TEÓRICO INTERNACIONAL 25
3.6. Medios de prevención y ayuda para la depresión 62
At Definition 9, we need to calculate the trajectory opportunistic vector between two parti-
cipants so we need to considerp1, p2 from the example in Figure 5.1. For each participant, we
need to calculate the trajectory vector (from definition 7)
T Vp1L1([8 : 00, 9 : 00), [9 : 00, 10 : 00), [10 : 00, 11 : 00), [11 : 00, 12 : 00), [12 : 00, 13 : 00), [13 : 00, 14 : 00)) = (0, 1, 1, 0, 0, 0) T Vp2L1([8 : 00, 9 : 00), [9 : 00, 10 : 00), [10 : 00, 11 : 00), [11 : 00, 12 : 00), [12 : 00, 13 : 00), [13 : 00, 14 : 00)) = (0, 1, 1, 0, 0, 0) then[Sp1p2(8 : 00, 9 : 00) = 0 [Sp1p2(9 : 00, 10 : 00) = 1 [Sp1p2(10 : 00, 11 : 00) = 1 [Sp1p2(11 : 00, 12 : 00) = 0
It is straightforward to calculate other calculations on different participants in the example.
5.3
Description of Mobility Tracking Data
This section focuses on the descriptive statistical analysis of the collected data set. The aim of this analysis is to provide a clear picture about the collected data before analysing it. Having a clear picture supports the understanding of the nature of the data and assists the analysis of the extracted opportunistic network.
5.3.1
Mobility Trajectory Profile
This type of data set represents the route of different users throughout the school buildings. For each route, detailed information was provided that includes: date and time of being at a spe- cific location (for more detailed information - see Chapter Three). Generally, this data set was highlighted to be able to extract the opportunistic network between different users. In order to analyse these data sets, they were visualized from different points of view as described in the following subsections. In short, 1609 mobility time intervals were collected during the experi- ment time slots. The time interval represents the presence of the participants for a continuous time period at the same place. In total, the mobility trajectories were collected for 160 hours through the experiment time slots.
70 5.3 Description of Mobility Tracking Data
User routine
Figure 5.2: Sum of Location Function per Day.
Figure 5.3: Radar Plot of Sum of Location Function per Day.
In this empirical study, the tracking of the participants started at 8:00 am and ended at 6:00 pm. This time periods reflects the existence of the participants in the school building according to the
5.3 Description of Mobility Tracking Data 71
school time table. Figure 5.2 collectively presents the different users’ routines by summing all
location functions (Definition 6) per day (i.e. P
P,LfLkpi(ts, te)). Although the first two weeks
were normal study weeks, there was no specific routine that the participants followed in their routine during their time at school. It was found that Tuesday and Friday in the first two weeks gained significant amounts of activity rather than week three. Mainly, participants were active from 10:00 am until 4:00 pm.
Figure 5.3 provides a different representation of users’ activities during every time slot in the experiment. The radar representation was used to represent the results as it represents periodic information about the users’ activities. It is very hard to represent periodic data using a linear plot as there would be a cutting point from the x-axis. This would affect the perception of various trends. However, it would also make the process of establishing any common pattern in their activities easier.
Venue Popularity
In this study, five different venues were strategically chosen as access points to monitor the
participants’ mobility. WhereL1 covers the one class room, lab and the south entrance. L2
covers two labs in the central building. L3 covers the North building the school office, and
part of the bridge between North building and Trevitech building. L4 covers the library at
second floor of the Trevitech building. L5 covers the Trevitech entrance and the refractory.
The topology of these venues is declared in Chapter Three. Regarding venue popularity, the participants were able to access different venues during all time slots of the experiments that
were aggregatedP
P,T fLkpi(ts, te) as seen in Figure 5.4. It is clear that participants’ activity
mainly focuses onL1 and L2 as they cover the most important class rooms and labs. In addition,
L4 did not gain a lot of access except for the last two days as this week was a revision week where some participants preferred to study in the Library.
72 5.3 Description of Mobility Tracking Data 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 5_11_2012 5_10_2012 5_09_2012 5_08_2012 5_04_2012 5_03_2012 5_01_2012 4_30_2012 4_27_2012 4_26_2012 4_24_2012 4_23_2012 L1 L2 L3 L4 L5
Figure 5.4: Venue Popularity Concerning Sum of Location Function per User.
Another visualization of the venues’ popularity in terms of trajectories (i.e Definition 7) and dur- ations (Definition 8) is seen respectively in Figures 5.5 and 5.6. The former presents the popular-
ity in terms of the number of trajectories for each venue (i.eP
P,TT VpiLk((t1, t2),· · · , (tn−1, tn)))
where the latter presents the popularity in terms of the total time intervals have been spent at
each venue (therefore P
P,T DVpiLk((t1, t2),· · · , (tn−1, tn))). Although the calculations are
different, the same result are produced so that L1, L2 are the most important (popular) venues
to the participants.
5.3 Description of Mobility Tracking Data 73
Figure 5.6: Venue Popularity in Terms of Durations .
Additionally, it has been found that the patterns seen in both Figures 5.5 and 5.6 are consistent with the occurrence of events for students.
5.3.2
Physical Interaction Profile
As mentioned in Chapter Three, the physical interaction file (system file) contains a number of hash codes. Each hash code represents the unique identification code for each wireless device (Bluetooth mote). This code was used instead of using the Bluetooth MAC address to overcome the limitation of the system file size of the Bluetooth mote. 2578 files were collected, with each file containing many codes with a maximum of 32 KB. Due to the features of this data set as well as the limitation of Bluetooth mote, it did not support the time of meeting, and so it is hard to visualize the data. As a result, the following subsection discusses the probability of missing encounters of the data set where encounter is the case where the Bluetooth devices become in range of each other (without any interaction) so that it can be detect by each other.
Probability of Missing Encounters
As the Bluetooth devices did not run simultaneously, there had to be missing data in recording some of the encounters. The following discussion calculates and clarifies the probability of missing encounters.