Database Reference
In-Depth Information
In Figure
8.4
f, we have applied
k
-means clustering to the 168 spatial situations
in terms of car presence and built a time mosaic display where each hourly
interval is represented by a square. As in the previous case, different colors have
been assigned to the clusters. The squares in the time mosaic are painted in
these colors. The squares are arranged so that the columns, from left to right,
correspond to the days, from Sunday (the first day in our data set) to Saturday,
and the rows correspond to the hours of the day, from 0 on the top to 23 at the
bottom. We see that the working days (Columns 2-6) have quite similar patterns
of coloring, which means similarity of the daily variations of the situations.
The patterns on Sunday (Column 1) and Saturday (Column 7) are different.
The multimap display in Figure
8.4
g shows summarized spatial situations: each
small map represents the mean presence values in the respective time cluster
(the color coding is the same as in the STC in Figure
8.4
c; see the legend in
the lower right corner). It is seen that the shades of cyan, which occur in the
night hours, correspond to very low car presence over the city and the shades of
red, which occur in the working days from 5 till 17 o'clock, to high presence,
especially on the belt roads around the city. Red also occurs in the afternoon of
Sunday (from 15 till 17) and in the morning of Saturday (from 8 till 9).
To deal with very large amounts of movement data, possibly not fitting in
RAM, discrete spatio-temporal aggregation can be done within a database or
data warehouse. The aggregates can then be loaded in RAM for visualization
and interactive analysis.
8.4.2 Tracing Flows
In the previous section, we have considered spatial aggregation of movement
data by locations (space compartments). Another method of spatial aggregation
is by pairs of locations: for two locations A and B, the moves (transitions) from
A to B are summarized. This can result in such aggregate attributes as number
of transitions, number of different objects that moved from A to B, statistics
of the speed, and transition duration. The term “flow” is often used to refer to
aggregated movements between locations. The respective amount of movement,
that is, count of moving objects or count of transitions, may be called “flow
magnitude.”
There are two possible ways to aggregate trajectories into flows. Assuming
that each trajectory represents a full trip of a moving object from some origin
to some destination, the trajectories can be aggregated by origin-destination
pairs, ignoring the intermediate locations. A well-known representation of the
resulting aggregates is the
origin-destination matrix
(
ODmatrix
) where the rows
and columns correspond to the locations and the cells contain aggregate values.
ODmatrices are often represented graphically as matrices with shaded or colored
cells. The rows and columns can be automatically or interactively reordered for
