Development of spatial econometric models for estimating the bicycle sharing trip activity

Abstract

In developing countries, the implementation of bicycle sharing programs is in a nascent stage; this paper provides a preliminary understanding of bicycle sharing trip activity introduced in Kerala, India. Fundamentally, a framework is utilized to incorporate spatial interaction effects and estimate the bicycle sharing trip activity using spatial econometric techniques. MYBYK is the associated bicycle rental company, with 67 stations in the various major locations at Cochin, Kalamasery, Choornikkarra, and Aluva cities in India. Information such as socio-demographic, bicycle sharing trip activity, weather, and built environment information have been collected in the study regions. Initially, spatiotemporal variation is analyzed to understand the bicycle-sharing trip activity. Then, spatial indicators are determined using Moran's I technique, and the study area's locational information is transmitted to the spatial weight matrix to illustrate the spatial neighbourhood structure. Finally, Trip Production and Trip Attraction models were developed for estimating the bicycle sharing trip activity using five spatial econometric techniques as Spatial Autoregressive (SAR) model, Spatially Lagged X model (SLX), Spatial Error Model (SEM), Spatial Durbin Error model (SDEM), and Spatial Durbin Model (SDM). The models are evaluated, and an appropriate model is found using the Bayesian approach. The findings indicate that SDM effectively estimates the bicycle sharing trip activity, and its spillover effects are also addressed for each spatial indicator variable.