Quantifying saturation point of Beijing bike-sharing market from environmental benefit: A data mining framework

Abstract

This study quantitatively estimates the carbon dioxide (CO2) emissions savings from ride records for passengers whose travel behavior shifted from polluting modes (public transport and private car) to bike-sharing in Beijing. We present a framework for examining how travel time, distance, purpose, frequency, weather, and demographics affect passenger usage and estimate environmental benefits. The framework comprises modules of association rules, density-based spatial clustering, random forest, and CO2 emission estimation. Our findings show that commuters with a trip distance of 1–2 km are more likely to change their behavior patterns. Therefore, more CO2 emission savings accrue in developed districts where residential density and employment rates are higher, than in central districts. Beijing saves 4322.38 kg CO2 per day. In contrast, four districts are oversupplied and have reached saturation points in the number of bikes. Implications for planners suggest that they will be able to better control the number of bikes launched.