Demand Forecasting of a Public Bike-Sharing System Reflecting Dynamic Spatial Data

Abstract

Bike sharing is booming in Korea as a leisure traffic mode, but its original purpose was to reduce traffic congestion. This study developed a demand forecasting model for bike sharing connected to a subway station. For accurate demand forecasting, we used various kinds of data to reflect the spatial distribution of travel demand. We used machine learning method as prediction model, and random forest had the best predictive result.