Database Reference
In-Depth Information
exploration starts in V-Analytics by removing incomplete data and artifacts, and
sending the identifiers of the candidate trajectories to SECONDO.InSECONDO
the user issues an STP query, and moves the result back to V-Analytics for
validation. The visualization in V-Analytics helps the human analyst in refining
the query parameters. It can take as many cycles as needed between SECONDO
and V-Analytics till the results are satisfactory.
The STP query can be written in SECONDO so that the result contains the time
intervals in which the pattern occurred. These can be interpreted as movement
events (m-events) in V-Analytics, so that the analysis procedures in the previous
section are applicable. For example, one is able to explore the percentage of
stepwise descents during one day, the percentage of missed approaches for each
airport, the temporal distribution of missed approaches for a given airport, and
so on.
12.7 Conclusions
In this chapter, we gave an overview of up-to-date research techniques to explore
and analyze trajectories. We detailed our motivations, gave the process we used
to build trajectory data set, and explained three trajectory exploration techniques
(direct manipulation, m-event, and MOD queries).
First, we introduced FromDaDy, a multidimensional visualization tool mak-
ing it possible to explore large sets of aircraft trajectories with direct manip-
ulation techniques. It uses a minimalist interface: a desktop with a matrix of
cells, and a dimension-to-visual variables connection tool. Its interactions are
also minimalist: brushing, picking, and dropping. Nevertheless the combination
of these interactions permits numerous functions: the creation and destruction
of working views, the initiation and refinement of selections, the filtering of data
sets, the application of Boolean operations. The cornerstone of FromDaDy is
the trajectory spreading across views with a simple brush/pick/drop paradigm.
With the incremental trajectory exploration and direct manipulation, the user
can discover the worthwhile requests for data sets. In a sense, the user explores
the data set, and at the same time, explores the request to perform.
Second, we detailed a generic procedure for analyzing mobility data that is
oriented to a class of problems where relevant places need to be determined
from the mobility data in order to study place-related patterns of events and
movements. The procedure includes: (1) extraction of relevant events from tra-
jectories by queries involving diverse instant, interval, and cumulative character-
istics of the movement and relations between the moving objects and elements
of the spatio-temporal context; (2) density-based clustering of the events by
spatial positions, temporal positions, movement directions and, possibly, other
attributes, which may be done in two stages for an effective removal of noise and
getting clear clusters; (3) spatio-temporal aggregation of events and trajectories
