Multi-Agent System for Demand Prediction and Trip Visualization in Bike Sharing Systems

Abstract

Featured Application: The main application of this work is the analysis and prediction of the demand in bike sharing systems using their open/private data. A multi agent system is proposed and a case study is conducted using the data of a bicycle sharing system from a middle size city.Abstract: This paper proposes a multi agent system that provides visualization and prediction tools for bike sharing systems (BSS). The presented multi-agent system includes an agent that performs data collection and cleaning processes, it is also capable of creating demand forecasting models for each bicycle station. Moreover, the architecture offers API (Application Programming Interface) services and provides a web application for visualization and forecasting. This work aims to make the system generic enough for it to be able to integrate data from different types of bike sharing systems. Thus, in future studies it will be possible to employ the proposed system in different types of bike sharing systems. This article contains a literature review, a section on the process of developing the system and the built-in prediction models. Moreover, a case study which validates the proposed system by implementing it in a public bicycle sharing system in Salamanca, called SalenBici. It also includes an outline of the results and conclusions, a discussion on the challenges encountered in this domain, as well as possibilities for future work.

Multi-Agent Systems

“…This Special Issue contributes with three works in the area of smart cities. The first work [32] proposed a multi-agent system that provides visualization and prediction tools for bike sharing systems. The proposed MAS includes an agent that performs data collection and cleaning processes, and it is also capable of creating demand forecasting models for each bicycle station.…”

Smart Bicycle-Sharing System Design for the Historical Peninsula of Istanbul

“…Network design and redesign issues are related to the topography, traffic and distribution of stations within the number of bicycles (Vogel et al, 2011). System-user balance management issues relate to user demand and satisfaction, while operational issues concentrate on the bicycle usage pattern (Lozano et al, 2018). Yang and colleagues (2016) has study on promoting user satisfaction by analyzing using pattern and parking condition with existing bike-sharing data.…”

“…Yang and colleagues (2016) has study on promoting user satisfaction by analyzing using pattern and parking condition with existing bike-sharing data. Lozano et al (2018) employs agent based modeling for predicting bicycle demand and Bortner et al (2015) utilize for predicting bicycle waiting duration with intend to dimensipn the optimum bicycle number in bicycle parks. Saltzman and Bradford (2016) implement agent-based modeling to determine cycling duration by evaluating the tendency of the user to walk after leaving the bicycle.…”

Characterizing Situations of Dock Overload in Bicycle Sharing Stations

“…Branch (ii) aims at forecasting the occupancy level of a station in the near future (i.e., with a time horizon between 30 min and 2 h ahead) by applying supervised machine learning techniques (e.g., regression [13][14][15]24], classification [16,17]). Based on these predictions, a recommender system can be integrated into the mobile application of the provider to suggest the stations close to the user-specified point of interest with a sufficient number of free docks/available bicycles.…”

“…Based on these predictions, a recommender system can be integrated into the mobile application of the provider to suggest the stations close to the user-specified point of interest with a sufficient number of free docks/available bicycles. Predictions are based not only on past occupancy levels but also on contextual information (e.g., meteorological data [24]). The main differences between the aforesaid works and the proposed approach are enumerated below: (i) Unlike the aforesaid approaches, this work does not address the problem of forecasting the station occupancy levels using supervised techniques.…”