Information Technology Reference
In-Depth Information
Fig. 3.9 Interactive data mining interface for mining flock patterns using decentralized movement
mining algorithms, implemented in REPAST ( P6 . Laube et al. 2008b ) (With kind permission from
Springer Science+BusinessMedia: Cova, T. J., Beard, K., Goodchild, M. F., and Frank, A. U., (eds.),
Geographic Information Science, Lecture Notes in Computer Science , vol. 5266, 2008, ISBN 978-
3-540-87472-0, Laube, P., Duckham, M., and Wolle, T., Decentralised movement pattern detection
amongst mobil geosensor nodes, pp. 199-216, Fig. 5)
Validation demonstrates that a method (a.k.a. data mining technique) within its
domain of applicability possesses a satisfactory range of accuracy consistent with
the intended application of the method. For instance, when suggesting a segmenta-
tion algorithm, validation could consist of cross-checking segments derived from
raw trajectory data with segments annotated by experts. Hence, validation can be
referred to as external evaluation.
Credibility refers to the degree of belief in the validity of a method (a.k.a. data min-
ing technique) to justify its use for research and decision making. The credibility
of a new method is relative to a particular application context. A credible method
is one that domain experts accept and consider suitable for a given problem.
Efficiency investigates the question if the proposed tool (a.k.a. data mining tech-
nique) can efficiently be used in an applied context. This includes the analysis
of algorithmic performance of the proposed methods and hence also investigates
their scalability.
Note the similarities to the qualities of patterns outlined in Sect. 3.1.1 .Thework
summarized in this topic repeatedly addresses the above concepts. In the following,
selected examples are singled out illustrating the concepts in the general context
of movement mining in computational movement analysis. Verification is a basic
Search WWH ::




Custom Search