CumInCAD is a Cumulative Index about publications in Computer Aided Architectural Design
supported by the sibling associations ACADIA, CAADRIA, eCAADe, SIGraDi, ASCAAD and CAAD futures

PDF papers
References

Hits 1 to 1 of 1

_id ddss2004_ra-19
id ddss2004_ra-19
authors Akamine, A. and A. NÚlson Rodrigues da Silva
year 2004
title An Evaluation of Neural Spatial Interaction Models Based on a Practical Application
source Van Leeuwen, J.P. and H.J.P. Timmermans (eds.) Recent Advances in Design & Decision Support Systems in Architecture and Urban Planning, Dordrecht: Kluwer Academic Publishers, ISBN: 1-4020-2408-8, p. 19-32
summary One of the serious problems faced by the Brazilian municipalities is the scarcity of resources for building education infrastructure. This asks for an optimal allocation of the available resources that includes, among other things, a rational spatial arrangement of the supply points (i.e., schools) in order to increase the demand coverage (i.e., students). If it is possible to foresee the regions where the demand is going to be concentrated, it is then possible to plan the location of new facilities and to assess the impact on the future level of service of the entire system. Considering that one of the consequences of the location-allocation process is the distribution of trips from demand points to supply points throughout the city, therefore affecting the overall intraurban accessibility conditions to essential services such as education, there is a strong need of models that planners can rely on to predict the future trip distribution patterns. As a result, the objective of this work was to evaluate the performance of Artificial Neural Networks (ANN) when applied to spatial interaction models, the so-called Neural Spatial Interaction Models. This was done in a practical context, in contrast to the more theoretical works commonly found in literature. The practical application showed that the neural spatial interaction model had different performances when compared to the traditional gravity models. In one case the neural models outperformed the gravity models, while on the other case it was just the opposite. The explanation for this may be in the data or in the ANN model formulation, as discussed in the conclusions.
keywords Artificial Neural Networks, Spatial Interaction Models, Education Infrastructure
series DDSS
last changed 2004/07/03 20:13

No more hits.

HOMELOGIN (you are user _anon_815986 from group guest) CUMINCAD Papers Powered by SciX Open Publishing Services 1.002