Trajectory Data Warehouses - Mobility Data - page 80

Presence f( ,)

Episode

Distance f( ,)

route m( )
distance
duration
avgSpeed
riskLevel

Velocity f( ,)

Figure 4.7 Combining the TDW in Figure 4.2 with the GeoPKDD TDW in Figure 4.5 ,the
latter seen as set of time-dependent fields, one for each measure.

For each district we sum the presence measure for all cells contained in the
district and such that the interval defined by start and end time of a trajectory
intersects January 1, 2010.

4.6.3 Representing the GeoPKDD TDW as Continuous Fields

The reader may have noticed that each measure of the TDW in Figure 4.5
defines a collection of time-dependent fields, one for each user profile. These
time-dependent fields are defined over spatio-temporal cells that have a fixed
granularity of, say, one square kilometer and one hour. We can produce a time-
dependent field for each measure by projecting out the Profile dimension and
aggregating the measure with the functions studied in Section 4.5 . For example,
from the presence measure in Figure 4.5 we can produce a time-dependent
field Presence by adding up the total presence by hour and square kilometer
for all profiles using one of the functions sum_s or sum_t . We can proceed
analogously with every measure in the cube, ending up with a collection of time-
dependent fields with the same granularity as that of the original TDW. Notice
that the functions defining these fields are stepwise ones, that is, the value of the
measure is constant in each spatio-temporal cell.
As shown in Figure 4.7 , using the approach above we can combine the TDW
of our running example in Figure 4.2 with the GeoPKDD TDW in Figure 4.5 .As
an example, consider the following query, which combines the field Presence
from the GeoPKDD TDW with our running example for discovering dense
traffic areas in residential zones.

Query 4.9. “For districts with more than 70%of residential use, give the average
presence of cars in January 21, 2012, at each point of the district.”

SELECT d.name, projPres
FROM District d, LandUse l, Presence p