Estimativa de matrizes de origem-destino a partir de registros de chamadas de telefonia celular

Abstract

This work presents a spatial method to partition geographic units given a set of cell phone towers. This geospatial method make easier to deal with multiple data layers. A cell phone call database from 2014 was used to estimate origindestination matrices by day, covering ordinary and extraordinary mobility behavior. The Census data of 2010 was used to adjust population expansion factors. The user’s residence inferring made possible the conception of two common matrices: Home-Based Others (HBO) and Non-Home Based(NHB). The population habits’ changing was detected on long holidays. The transportation database was used to validate travels counting, and the results has showed to be compatible with the official estimations. The simple population expansion factor seems to overestimate the travels counting, mainly for regions with less data, i.e., better inferences come out when more cell phone data are available for the study area.