%0 Conference Proceedings
%4 sid.inpe.br/mtc-m18/2013/12.10.18.20
%2 sid.inpe.br/mtc-m18/2013/12.10.18.20.01
%@issn 2179-4820
%T Towards efficient prospective detection of multiple spatio-temporal clusters
%D 2013
%A Veloso, Bráulio,
%A Iabrudi, Andréa,
%A Correa, Thais,
%@affiliation Universidade Federal de Ouro Preto (UFOP)
%@affiliation Universidade Federal de Ouro Preto (UFOP)
%@affiliation Universidade Federal de Ouro Preto (UFOP)
%E Andrade, Pedro Ribeiro,
%E Santanchè, André,
%B Simpósio Brasileiro de Geoinformática, 14 (GEOINFO).
%C Campos do Jordão
%8 24-27 nov. 2013
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 12
%S Anais
%X In this paper we propose a novel technique to efficiently detect multiple emergent clusters in a space-time point process. Emergent cluster detection in large datasets is a ubiquitous task in any application area where fast response is crucial, like epidemic surveillance, criminology or social networks behavior changing. Although different authors investigate aspects of efficient spatio-temporal cluster detection, they handle either multiple or prospective detection of spatio-temporal clusters. Our work concomitantly presents a solution for both aspects: prospective and multiple cluster efficient detection in space and time. Our results with synthetic data are very encouraging, since with a wide range of parameters, we are able to detect multiple clusters in about 90% of the scenarios with very low type I and II errors (less than 2%), without increasing delay time.
%@language en
%3 paper7.pdf