Avaliação de modelos de regressão espacial para análise de cenários do Transporte Rodoviário

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

The Road Freight Transportation (RFT) currently presents itself as one of the most used ways of transportation in Brazil, representing 61.1% of the transportation matrix, against 20.7% of railroad, 13.6% of the river-way, 4.2% of duct-way and 0.4% of the aerial. The possibility of delivery at the door of the purchaser, as well as the easiness of loading and unloading of merchandises are some of the characteristics that corroborate its participation in the market. Therefore, the knowledge concerning the characteristics of the RFT, as well as the identification of the factors that influence its spatial distribution assist the interested parties in decision taking. The spatial distribution of the road freight fleet vehicles generally depends on regional characteristics, such as the production of grains, the existence of a port and so forth, indicating that the "space", in this case, the geographic one, influences directly in such distribution. Thus, conventional regression models are not adjusted to work with spatial data, where the spatial dependence is present, making necessary the use of spatial regression models. The present work shows, through a methodology for the construction of spatial regression models, an evaluation of them when compared with the conventional regression models, aiming at identifying and analyzing the intervening factors in the trucks fleet spatial distribution. For this, the road freight vehicles of the type granary ship with 2 or 3 axles and the simple trucks, also with 2 or 3 axles, was studied where the independent variables used were demographic density, exportations value, investments value, soy produced amount and industrials establishments amount. The results showed that, there being spatial dependence, the spatial regression models presented diagnoses that allow to understand the reality better than the conventional regression models, and that the form of the neighborhoods matrix influences directly in such diagnoses. It is concluded from the work that models that incorporate the random factor "space" supply better results to a more efficient planning.

ASSUNTO(S)

transporte rodoviário de carga vias de transporte regressão espacial evaluation of spatial regression models to analysis of scenes of the road freight transportation

Documentos Relacionados