Authors: Paula Simões; Sílvia Shrubsall; Isabel Natário
Addresses: Instituto Superior de Engenharia de Lisboa (IPL) and CMA, Rua Conselheiro Emídio Navarro, 1, 1959-007 Lisboa, Portugal ' Centro de Sistemas Urbanos e Regionaís (CESUR), Instituto Superior Técnico, Avenida Rovisco Pais, 1 – 1049-001 Lisboa, Portugal ' Faculdade de Ciências e Tecnologia (UNL) and CEAUL, Quinta da Torre, 2825-114 Caparica, Portugal
Abstract: This paper presents a spatial econometrics analysis for the number of road accidents with victims in the smallest administrative divisions of Lisbon, considering as a baseline a log-Poisson model for environmental factors. Spatial correlation is investigated for data alone and for the residuals of the baseline model without and with spatial-autocorrelated and spatial-lagged terms, considering transformed data to meet the specificities of the application of these techniques. In all the cases no spatial autocorrelation was detected. Given the ongoing analysis and the discrete nature of data, several hierarchical log-Poisson models were further fitted, in a Bayesian setting, implementing a different approach and finding some evidences of spatial structure in data.
Keywords: spatial econometrics; Moran's I; spatial autoregressive model; SAR model; spatial error model; SEM; Lagrange multipliers tests; hierarchical log-Poisson Bayesian model; conditional autoregressive priors; road accidents; Portugal; traffic accidents; Lisbon.
International Journal of Business Intelligence and Data Mining, 2015 Vol.10 No.2, pp.152 - 173
Available online: 06 May 2015 *Full-text access for editors Access for subscribers Purchase this article Comment on this article