Station-level demand prediction in bike-sharing systems through machine learning and deep learning methods

Abstract

Public Bike-Sharing systems have been employed in many cities around the globe. Shared bikes are an efficient and convenient means of transportation in advanced societies. Nonetheless, station planning and local bike-sharing network effectiveness can be challenging. This study aims to address these challenges by predicting the station-level demand in public bicycle-sharing systems through the application of machine learning and deep learning methods. This has been done by outlining the influencing urban built environment factors of bike-sharing systems and comparing the machine and deep learning models’ performance prediction results.
The Spatial Regression Graph Convolutional Neural Network (SRGCNN) and SRGCNN-Geographically Weighted methods are employed and compared with traditional machine learning methods, namely Multiple Linear Regression, Multilayer Perceptron Regressor, Support Vector Machine, and Random Forest Regressor. For the SRGCNN method, a graph has been constructed based on the number of trips between the stations. Additionally, factors related to the urban environment have been considered for the prediction task. Also, evaluation metrics have been utilized to assess and facilitate the comparison of the machine and deep learning models’ performance.
Concluding this case study in Zurich, the deep learning models and Random Forest Regressor were found to outperform other machine learning methods, showcasing better accuracy in terms of several evaluation metrics. Despite recognizing limitations such as the relatively low number of stations, this study highlights future research opportunities to enhance model accuracy and deepen our understanding of micro-mobility demand dynamics, especially in the context of PBS station planning.

Popular Abstract

The field of artificial intelligence has drawn much attention in the scientific community. Its integration into the field of Geographic Information Systems has given a new perspective on solving spatial-related problems. Micro-mobility planning can be pretty challenging since it comes alongside lots of data and requires computational power, sufficient storage space, and data processing. This is where Geographic Artificial Intelligence facilitates geospatial data analysis, processing, and making predictions to assist policymakers and urban and transport planners in the case of this study.
Public-Bicycle Sharing systems have been developing in many cities around the world. It is an efficient and convenient means of transport in the citizens' daily lives. However, many questions arise regarding station placement, public bike availability across cities, and more. This thesis aims to tackle those issues by predicting the station-level demand in the bike-sharing system in Zurich.
Publicly available data have been used to calculate the number of shared bikes at each station for approximately one month. Additionally, other factors, such as healthcare centers, sports facilities, the closest distance from each station to public transport stops, and more, have been taken into consideration since they can significantly impact the usage of shared bikes. Subsequently, machine and deep learning models have been developed to predict the station-level demand for the bike-sharing system in Zurich. After developing the models, evaluation metrics have been utilized to compare the models’ performance results.
Concluding the case study in Zurich, Geographic Artificial Intelligence proved to be useful in predicting the station-level demand for bike-sharing systems. Despite acknowledging limitations, such as the relatively low number of stations, this study highlights the capabilities of such models and their potential application in addressing micro-mobility problems, particularly in station planning, across different cities worldwide.

• author: Staikos, Nikolaos
• supervisor: Pengxiang Zhao; Ali Mansourian
• organization: Dept of Physical Geography and Ecosystem Science
• course: NGEM01 20231
• year: 2024
• type: H2 - Master's Degree (Two Years)
• subject: Earth and Environmental Sciences
• keywords: Physical Geography, Ecosystem Analysis, Bike-sharing demand, Machine learning, Deep learning, Spatial regression, Graph Convolutional Neural Network, Multiple Linear Regression, Multilayer Perceptron Regressor, Support Vector Machine, Random Forest Regressor, Urban environment, Micro-mobility, Station planning, Geomatics
• publication/series: Student thesis series INES
• report number: 634
• language: English
• id: 9146401
• date added to LUP: 2024-01-23 09:55:57
• date last changed: 2024-01-23 09:55:57