Determinants of paratransit feeder service provision using AI-synthesized revealed preference data: a paratransit drivers’ perspective on PT reform using machine learning

Document Type

Journal Article

Publication Date

2025

Subject Area

mode - bus, mode - paratransit, ridership - drivers, operations - coordination, planning - surveys, planning - methods

Keywords

Revealed preference, feeder-trunk, public transport, paratransit, machine learning, artificial neural network

Abstract

Although there is a wealth of research on passengers’ perception of public transport services and their willingness to use them, only a few have examined paratransit drivers’ intention to provide service. In this study, we have proposed using combined forward sequential feature selection with random forest classifier (FSFS-RFC) and Artificial Neural Network (ANN) to investigate the factors influencing drivers’ intention to provide feeder services to a dedicated bus lane system in Freetown. Focus group discussion was held with heads of various public transport unions, and a questionnaire survey was administered to a sample of 1124 paratransit drivers. The proposed FSFS-RFC and ANN algorithms revealed consistent results with schemes for fleet renewal, provision of coordinated paratransit feeder terminals, guaranteed hours of bus operation and increase in daily profit as the most significant determinants. The results imply that the algorithms can learn from AI-synthesized data on paratransit drivers’ operations for efficient planning.

Rights

Permission to publish the abstract has been given by Taylor&Francis, copyright remains with them.

Share

COinS