Multilayer Feedforward Networks for Transportation Mode Choice Analysis: An Analysis and a Comparison with Random Utility Models

Although discrete choice analysis is usually based on random utility theory, a different approach to choice analysis, based on artificial neural network (actually a multilayer feedforward network—MLFFN) models, recently has been proposed. This modeling approach can address three main demand simulation issues (trip generation, trip distribution and modal split) and has shown good predictive capability. Most of the research using this approach deal with extra-urban (inter-city or intra-regional) trips and they are calibrated on aggregate data, simulating demand flows. An alternative approach can be followed up by using disaggregate data, as only one paper has done, to simulate single-user choice. The aim of the current paper is to first describe the main step towards the successful application of MLFFNs to support travel demand analysis, and then to show that they can be fruitfully applied to analyze transportation mode choice. A deep analysis has been carried out to address each of the major issues needed to make an MLFFN operational. The proposed approach relies on disaggregated (revealed preferences survey data-sets) data taken from two different case studies. The two case studies focus on medium distance intercity journeys, and allow the investigation of mode choice for two different trip purposes and two different geographical contexts: journey-to-work of commuters within the Italian region of Veneto and journey-to-study of students towards the rural location of the University of Salerno. MLFFNs performances have been compared with the most effective and advanced closed-form random utility models (RUMs) that can be calibrated on the same survey data-sets. The models are validated and compared using indices common to MLFFN or RUM applications. Results demonstrate that MLFFNs can be a feasible tool for travel demand analysis.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01002071
  • Record Type: Publication
  • Files: TRIS, ATRI
  • Created Date: Jul 14 2005 9:20PM