|
Article - Student Track Prevaba: a Bayesian Model to Predict the Existence of Victims in car accidents Author: TELLES, M.J., SANTINI, P.H., dos SANTOS, J.V.C., BARBOSA, J. L. V. Abstract: Road safety is an area which is concerned with both the reduction of accidents as with the care provided to the victims. Several initiatives are proposed to assist with reducing the number of accidents, such as surveillance, awareness campaigns and support equipment to drivers. Other initiatives for prevention and protection are proposed by vehicle manufacturers in terms of requirements of governament entities. As a final resort, that is, in the event of the accident and the victim needs medical attention, this should be done as quickly as possible. To assist in identifying the existence of the victim and the need for medical care, we propose a Bayesian model, called Prevaba, which uses Bayesian Networks (BN), which aims to predict the existence of victims in traffic accidents. In order to validate the model, we developed a prototype that performed the actual data classification in Porto Alegre - RS for the year 2013. The prototype made the classification based on the previous year's data (2012), showing an index above 90% accuracy, taking into account the incorrect classi cations are only classified as victimless, but actually was has a victim. Keywords: Statistical inference, classifiers, data mining, decision making. Complete Reference: Telles, M.J., Santini, P.H., dos Santos, J.V.C., Barbosa, J. L. V., "Prevaba: Um Modelo Bayesiano para Predição da Existência de Vítimas em Acidentes de Trânsito", Revista de Sistemas de Informação da FSMA n 16(2015) pp. 16-25 |