Skip to main content
CISTI'2017 - 12ª Conferência Ibérica de Sistemas e Tecnologias de Informação

Full Program »

Analysis of Road-Traffic Behavior In The Metropolitan Area of Recife-Brazil

Federal highways that cross the Metropolitan Regions of some cities are constantly congested, not only by the number of vehicles, but also because they are subject to stoppages of the most diverse natures, such as accidents, holes, natural weather and other types of problems. We propose a classification model of behavioral patterns for the federal highways that cross the metropolitan area of Pernambuco, which is a state in the Northeastern region of Brazil. The proposed model allows some events anticipation, especially those that may cause constraints, retention, or reduction of traffic flow. The data source of this research is from the database of the Federal Highway Police of Pernambuco (PRF) since 2007 until 2015. We have considered vehicles, track layout and road sections related to accidents, among others. Based on the information obtained, a Data Mining was performed using the CRISP-DM methodology to find behavioral patterns on highways and in their surroundings. Machine learning algorithms were used for classification and regression, being prioritized, Decision Trees and Neural Networks. The values of the area under the ROC (AUC) curve obtained were above 0.7 which reflects a good degree of reliability. The proposed prediction model means an advance in terms of mobility and cargo transport management, since it allows anticipating events and behaviors, favoring the choice of alternative routes and increasing the time for choice for certain routes.

Author(s):

Othon Oliveira    
Universidade de Pernambuco
Brazil

Fernando Lima Neto    
Universidade de Pernambuco
Brazil

 

Powered by OpenConf®
Copyright ©2002-2016 Zakon Group LLC