Detecção de clusters espaciais em modelos de regressão beta

Abstract

Spatial Scan Statistics has been developed for geographical cluster detection in different types of models, for example, Bernoulli, Multinomial, Poisson, Exponential, Weibul and Normal. However, some data are continuos in the interval (0;1) such as rates and proportions or are limited in the interval (a;b), a < b. Therefore, in this work, we propose a spatial scan statistic for Beta regression model. The test statistics is based on a likelihood ratio test and evaluated using Bootstrap p-value. The proposed method is illustrated using infant mortality in the Amazonas State, Brazil. The Statistical power, sensitivity and positive predicted value of the test are examined through a simulation study.