Detec？？o de agrupamentos espa？o-temporais para identifica？？o de áreas de risco de homicídios por arma branca em Jo？o Pessoa, PB

scan statistic has been generally used to identify spatial clusters derived from adverse phenomena, as well as from epidemiological studies. poisson and space-time permutation models are appropriate for counting data, where the null hypothesis assumes randomness in a space-time distribution of a phenomenon. in space-time studies randomness is seen as the same occurrence risk in every location, regarding (expected) time period. for analysis purposes, it was considered homicide data occurred in the jo？o pessoa city, paraíba state, brazil, by stabbing between 2001 and 2009. from these data, possible patterns in homicide occurrence were investigated and therefore, space-time behavior of the phenomenon was analyzed, identifying high risk areas. after comparing maps using an epidemiological approach, it was verified that scan statistic using poisson model, featuring 7% of the population restriction, had the best results. the model identified the presence of frequent clusters (recidivist clusters) over the years in the northern and western regions of the city.