Premium
Discovery of anomalous spatio‐temporal windows using discretized spatio‐temporal scan statistics
Author(s) -
Janeja Vandana P.,
Gangopadhyay Aryya,
Mohammadi Seyed,
Palanisamy Revathi
Publication year - 2011
Publication title -
statistical analysis and data mining: the asa data science journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.381
H-Index - 33
eISSN - 1932-1872
pISSN - 1932-1864
DOI - 10.1002/sam.10122
Subject(s) - computer science , data mining , window (computing) , morphing , spatial analysis , classification of discontinuities , artificial intelligence , statistics , mathematics , mathematical analysis , operating system
In this paper, we address the discovery of anomalous spatio‐temporal windows using discretized spatio‐temporal scan (DSTS) Statistics. Anomalous spatio‐temporal window discovery is required in several key applications such as disease outbreaks in a region over a period of time, monitoring drinking water quality over time, identifying health risks to the population in a polluted region and urbanization patterns in a city, to name a few. In this paper, we address the issues arising out of the simultaneous effects of the properties of space and time in the discovery of anomalous windows. In such a framework, we identify (i) at what point in time the window changes, (ii) the spatial patterns of change over time, and (iii) a spatial extent in time which is completely or partially deviant with respect to the rest of the anomalous spatio‐temporal windows. None of the current approaches address all these issues in combination. We identify this knowledge keeping in mind the spatial and temporal autocorrelation, morphing shape of the window, and possible spatial or temporal discontinuities of the window. Subsequently, we perform experiments on several real‐world datasets, to validate our approach, while comparing with the established approaches. © 2011 Wiley Periodicals, Inc. Statistical Analysis and Data Mining 2011