Dynamic factor modelling aimed to identify the evolution of the economic activity has been widely used in several industrialized countries. Though the high adaptability of such methodologies to scarcity information contexts, results may present a lack of robustness. The purpose of this paper is to prove empirically that in regional contexts, it is possible to obtain a better fitting by means of the utilization of the optimization models with binary variables. In this way, it is possible to capture the whole relationship that exists between a subset of partials indicators and the reference series.