Abstract
The value of neural network analysis as a tool in transportation planning is tested on two transportation issues, one behavioural and the other physical. The behavioural case involved travel behaviour forecasting, comparing trave demand patterns of men and women in Israel, to determine the connection between them and a variety of key socio-economic and demographic variables. The physical case addressed a problem of traffic control at a major intersection in Ramat Aviv, Israel. Both involve the kind of complex, highly dimensional, large scale data to which the methodology has been applied successfully in the physical sciences and in cognitive modelling, but is relatively new in the social sciences. Neural network methodology is defined, a variety of application categories presented to explain the rationale for the ones chosen, and both experiments detailed. The neural network results were more impressive for the physical application than for the behavioural study because of deficiencies in the data for the latter. Recommendations are made for the kind of information that would enrich the data and permit neural networks to be utilized to full advantage.
| Original language | English |
|---|---|
| Pages (from-to) | 141-204 |
| Number of pages | 64 |
| Journal | Progress in Planning |
| Volume | 50 |
| Issue number | 3 |
| DOIs | |
| State | Published - Oct 1998 |
Bibliographical note
Funding Information:The intent of this study was to demonstrate the strengths and weaknesses of neural network modelling on the kind of data usually available to transportation planners. For the behaviour application in the Israeli context, this means within the framework of the 1984 Traveling Habits Survey ( Central Bureau of Statistics, 1987 ). This is the most recent travel survey commissioned and financed by the Ministry of Transport. The survey's content and format are similar to the prevalent American and European surveys currently intended to serve transportation planners. While out-of-date for practical current planning, it serves the purpose as a theoretical example of applying neural network analysis.
Funding Information:
Many thanks to Etti Dolev and Paul Feigin of the Technion's Statistics Laboratory for their help with the statistical analysis, and Ayelet Gal-Tsur of the Technion's Centre for Transportation Research for running the NETSIM simulation. The researching of the travel behaviour case study was supported by the Samuel Strassler Memorial Fellowship and the Technion's Graduate School, and carried out under the supervision of Daniel Shefer and Ilan Salamon. I am grateful to both of them for their guidance, comments and advice. Finally, I thank Derek Diamond for his encouragement and patience.
ASJC Scopus subject areas
- Geography, Planning and Development