Sistema de Información Científica Redalyc
Red de Revistas Científicas de América Latina y el Caribe, España y Portugal
In this paper, we propose a strategy to improvethe forecasting of traffic accidents in Concepci ́on, Chile.The forecasting strategy consists of four stages: embedding,decomposition, estimation and recomposition. At the first stage,the Hankel matrix is used to embed the original time series. At thesecond stage, the Singular Value Decomposition (SVD) techniqueis applied. SVD extracts the singular values and the singularvectors, which are used to obtain the components of low andhigh frequency. At the third stage, the estimation is implementedwith an Autoregressive Neural Network (ANN) based on ParticleSwarm Optimization (PSO). The final stage is recomposition,where the forecasted value is obtained. The results are comparedwith the values given by the conventional forecasting process. Ourstrategy shows high accuracy and is superior to the conventionalprocess.

Palabras clave: Autoregressive neural network, particle swarm optimization, singular value decomposition.
Ver Resumen
Universidad Autónoma del Estado de México
Sistema de Información Científica Redalyc ®
Versión 3.0 | 2017