Traffic Accidents Forecasting using Singular Value Decomposition and an Autoregressive Neural Network Based on PSO

In this paper, we propose a strategy to improve the 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 the second stage, the Singular Value Decomposition (SVD) technique is applied. SVD extracts the singular values and the singular vectors, which are used to obtain the components of low and high frequency. At the third stage, the estimation is implemented with an Autoregressive Neural Network (ANN) based on Particle Swarm Optimization (PSO). The final stage is recomposition, where the forecasted value is obtained. The results are compared with the values given by the conventional forecasting process. Our strategy shows high accuracy and is superior to the conventional process.

[1]  M. Mohandes.  Modeling global solar radiation using Particle Swarm Optimization (PSO) , 2012 .

[2]  Daniel Svozil,et al.  Introduction to multi-layer feed-forward neural networks , 1997 .

[3]  Taehoon Hong,et al.  An estimation model for determining the annual energy cost budget in educational facilities using SARIMA (seasonal autoregressive integrated moving average) and ANN (artificial neural network) , 2014 .

[4]  T. Shores.  Applied Linear Algebra And Matrix Analysis , 1999 .

[5]  Seok-Beom Roh,et al.  Design of fuzzy radial basis function-based polynomial neural networks , 2011, Fuzzy Sets Syst..

[6]  Goutami Chattopadhyay,et al.  Autoregressive forecast of monthly total ozone concentration: A neurocomputing approach , 2009, Comput. Geosci..

[7]  Adel Al-Jumaily,et al.  Multi Neural Networks Investigation based Sleep Apnea Prediction , 2013 .

[8]  Yohsuke Kinouchi,et al.  Neural networks for event extraction from time series: a back propagation algorithm approach , 2005, Future Gener. Comput. Syst..

[9]  Yong Li,et al.  PSO-based neural network optimization and its utilization in a boring machine , 2006 .

[10]  Ah Chung Tsoi,et al.  Universal Approximation Using Feedforward Neural Networks: A Survey of Some Existing Methods, and Some New Results , 1998, Neural Networks.

[11]  Luis Fernando de Mingo López,et al.  The optimal combination: Grammatical swarm, particle swarm optimization and neural networks , 2012, J. Comput. Sci..

[12]  Héctor Pomares,et al.  Time series analysis using normalized PG-RBF network with regression weights , 2002, Neurocomputing.

[13]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[14]  Asaad Y. Shamseldin,et al.  Comparative study of different wavelet based neural network models for rainfall–runoff modeling , 2014 .

[15]  Leslie S. Smith,et al.  A novel neural network ensemble architecture for time series forecasting , 2011, Neurocomputing.

[16]  Mu-Chen Chen,et al.  Forecasting the short-term metro passenger flow with empirical mode decomposition and neural networks , 2012 .

[17]  Mauro Birattari,et al.  Swarm Intelligence , 2012, Lecture Notes in Computer Science.

[18]  David M. Skapura,et al.  Neural networks - algorithms, applications, and programming techniques , 1991, Computation and neural systems series.

[19]  Hiok Chai Quek,et al.  RLDDE: A novel reinforcement learning-based dimension and delay estimator for neural networks in time series prediction , 2007, Neurocomputing.

[20]  Mehdi Khashei,et al.  Hybridization of autoregressive integrated moving average (ARIMA) with probabilistic neural networks (PNNs) , 2012, Comput. Ind. Eng..