Transit Shift Response Analysis Through Fuzzy Rule Based-Choice Model: A Case Study of Indian Metropolitan City

Autor: Kedia, Ashu, Sowjanya, D., Salini, P., Jabeena, M., Katti, Bhimaji
Zdroj: Transportation in Developing Economies; April 2017, Vol. 3 Issue: 1 p1-13, 13p
Abstrakt: The productivity of most of the public transport services has greatly reduced due to the absence of appropriate transit planning strategies. The efficacy of public transport system is undoubtedly governed by transit accessibility, and provision of transit supply and service facilities by transport organizations. If a transit system, and its operations prove to be inefficient, its patronage shifts towards private mode or para-transit, resulting in an uneconomical and environmentally degraded system. Therefore, the identification of attributes influencing consumers’ choice of transit system becomes an important aspect in urban context. The conventional Logit modelling approach, based on the theory of random utility maximization has been attempted by many. Moreover, the approach possesses certain limitations in addressing the uncertainty lying in humans’ choice decisions and ambiguous expressions made by travellers for the available options. Whereas, Fuzzy Logic, an Artificial Intelligence technique works on the principle of simple and logical ‘If-Then’ rules, framed to predict the preferred choice through approximate reasoning. The study precisely attempts to analyze the transit choice behaviour of urbanites, with the help of fuzzy rule based transit choice model, considering Surat, a fast developing Indian metropolitan city in the state of Gujarat as the study area. Walking distance to a bus stop, waiting time at a bus stop, and bus schedule reliability are observed to be the governing accessibility attributes, and thus are considered further for model development. Furthermore, sensitivity analysis showing the impacts of accessibility attributes on transit shift is carried out as an application of the developed model.
Databáze: Supplemental Index