Modeling bike availability in a bike-sharing system using machine learning

TL;DR

This paper compares machine learning models for predicting bike availability in a bike-sharing network, finding univariate models more accurate but multivariate models useful for spatially correlated stations, with a 15-minute prediction horizon being optimal.

Contribution

It introduces a multivariate regression approach using PLSR to account for spatial correlations in bike-sharing stations, enhancing prediction modeling.

Findings

Univariate models outperform multivariate models in prediction accuracy.

Station neighbors and 15-minute horizon significantly improve predictions.

Multivariate models are suitable for large, spatially correlated networks.

Abstract

This paper models the availability of bikes at San Francisco Bay Area Bike Share stations using machine learning algorithms. Random Forest (RF) and Least-Squares Boosting (LSBoost) were used as univariate regression algorithms, and Partial Least-Squares Regression (PLSR) was applied as a multivariate regression algorithm. The univariate models were used to model the number of available bikes at each station. PLSR was applied to reduce the number of required prediction models and reflect the spatial correlation between stations in the network. Results clearly show that univariate models have lower error predictions than the multivariate model. However, the multivariate model results are reasonable for networks with a relatively large number of spatially correlated stations. Results also show that station neighbors and the prediction horizon time are significant predictors. The most effective prediction horizon time that produced the least prediction error was 15 minutes.