Previsão de demanda na fase de planejamento antecipado de projetos de transporte de passageiros: uma abordagem por redes neurais artificiais

Abstract

Considering that passenger demand is a major risk in passenger rail infrastructure projects, this study aims to validate a demand forecasting model based on artificial neural networks (ANN), in order to contribute to the project management of this type of projects, it is still in front-end planning of these projects. For this, the design of the type ex-post facto was used in a descriptive research with quantitative approach where the research group was formed by subway and train stations in the metropolitan region of São Paulo (RMSP). The data for training, testing and validation of the neural model demand forecast were obtained from secondary sources, which are: the Urban Mobility Research 2012 in the RMSP; and the data base of entry passenger at subway and train stations. Proposed were 12 architectures of the ANN with 15 different configurations, totaling 180 training processes, testing and validation. For each of the architectures, the lowest mean square error (MSE) obtained was identified; and the best architecture, with a hidden layer was performed relevance analysis by Garson method, the model 4 input variables: the population; the school enrollment; the number of jobs; and per capita income. The main results of this study demonstrate the validity of the proposed architectures, presenting MSE% from 0.045% ~ 0.109%. The practical contribution this study is to serve as an aid tool for organizations and project managers in the study of economic and financial viability of these projects, still in its early planning stages, serving as an investment decision-making tool.