Abstract

Road accidents are now a major health problem worldwide, and particularly in Angola, where they are currently one of the major causes of death in the country. Over the time horizon under study, from 2002 to 2015, the average growth rate of road accidents was 7.3% and for both deaths and injured was 11.7%. In the present work, we have characterize the trend of the road accidents, deaths and injured in Angola using an STL, Seasonal- Trend decomposition by Loess. Time series modeling techniques were used to estimate the mathematical model that best fits the original data, in order to explain the evolution of the series and to make predictions. We have used classic Seasonal ARIMA models in two different approaches, the first one treat all observations the same way. The second approach identifies outliers, taking into account its magnitude and estimates Seasonal ARIMA models for the series excluding the significant outliers. The most appropriate models (in terms of the usual validation criteria) were identified for the characterization of the time series variability relative to road accidents, deaths and injured in Angola, as well as for providing the best predictions.