Data Analysis: Space-Time Clustering

1615 Words4 Pages

DEFINITION
Space-Time Clustering is a method of data analysis whereby the data objects are grouped with reference to a specific place and time. Space time clustering is therefore involve finding clusters that emerge during a particular time interval at particular places. Spatial cluster detection allows the identification of locations, shapes and sizes of potential anomalous spatial regions. The analysis of these clusters aid in the understanding of current patterns and prediction of future ones using a data-driven approach or a model-based approach.

SYNONYMS: Spatial surveillance, Space-Time Scan Statistic, Analysis Tab, Spatial Window Tab, Temporal Window Tab

BACKGROUND
The goal of space-time cluster detection is to identify the location, shapes and sizes of potentially anomalous spatial regions and analyze these clusters to determine if they are contingent clusters or legit cluster portraying some unforeseen information. Space-time cluster analysis therefore leads to the establishment of the characteristics, scale, scope and detection of various phenomena.

METHODOLOGY
Statistical local measures space-time clustering can be achieved using space-time scan statistic; while global space-time clustering is possible using algorithms like the Knox index, mantel test and space-time K-function.

Space-time clustering involves a spatio-temporal scan of a space with a 3 dimensional search window that independently, but simultaneously captures the space and time. The procedure involved in spatial scan can be summarized as follows:
Generate a model under the null hypothesis (no clusters) and under an alternative hypothesis.
Derive a score function using the null hypothesis and the alternative hypothesis (hypothesis gotten using the...

... middle of paper ...

...ed areas. When we want to study a change in pattern or frequency, Cluster and Outlier Analysis could prove effective. Grouping Analysis could be valuable in studying characteristics of individual patterns, e.g. disease outbreaks.

REFERENCES
Assuncao, Renato, et al. "Space-time cluster identification in point processes." The Canadian Journal of Statistics. 2007. 9-25.
Cromley, Ellen k. and Sara McLafferty. GIS and Public Health. Guilford Press, 2011.
Gould, MS, S Wallenstein and M Kleinman. "Time-space clustering of teenage suicide." PubMed. 1990.
Kulldorff , Martin , et al. "A space-time permutation scan statistic for disease outbreak detection." PLOS Medicine. 2005.
Wilson , Margo and Martin Daly. "Spatial-Temporal Clustering of Chicago Homicides." Proceedings of the 4th Annual Symposium of the Homicide Research Working Group. Washington, DC, 1997. 160-163.

Open Document