The forecasting of passenger demand under hybrid ridesharing service modes: A combined model based on WT-FCBF-LSTM. (November 2020)

• Record Type: Journal Article
• Authors: Li, Xuefeng; Zhang, Yong; Du, Mingyang; Yang, Jingzong
• Is Part Of: Sustainable cities and society. Volume 62(2020)
• Publication Date: 2020-11
• Subjects: Express service -- Ridespliting service -- Spatial-temporal characteristics -- Combined model -- Demand forecasting; Sustainable urban development -- Periodicals; Sustainable buildings -- Periodicals; Urban ecology (Sociology) -- Periodicals; 307.76
• DOI: 10.1016/j.scs.2020.102419
• Languages: English
• ISSNs: 2210-6707

Abstract

Highlights: Spatiotemporal characteristics of express and ridespliting services are analyzed. Factors affecting the demands of these two services are examined. A combined WT-FCBF-LSTM model is proposed to predict the passenger demand. The distribution of ridespliting demand is concentrated and the peak is obvious. WT-FCBF-LSTM can well capture different characteristics of these two services. Abstract: In order to predict the passenger demand under hybrid ridesharing service modes, firstly, based on the order data of DiDi Chuxing in Haikou, China, the spatial-temporal characteristics of the demands for express and ridespliting services are compared and analyzed, and the influential factors of these two modes' passenger demands are identified. Then, considering the historical order demand, travel time rate, the demand of neighbouring regions, day-of-week, time-of-day, weather and points of interest, a combined model based on WT-FCBF-LSTM (Wavelet Transform, Fast Correlation-basd Filter, and Long Short-term Memory) is proposed to predict the passenger demand in different regions for different time intervals. Finally, the parameter tuning and validity analysis for the combined model are carried out. The results show that the peak of wave for ridespliting demand is more obvious than that of express demand in the morning and evening peak periods, and ridespliting service has a certain market potential in urban transportation hubs. Compared with LSTM, WT_LSTM and FCBF_LSTM models, WT-FCBF-LSTM can improve the prediction accuracy and well capture the different spatial-temporal characteristics of express and ridespliting services.