Citation

Ma Z, Zhang P. Multimodal Transp. 2022; 1(1): e100002.

Copyright

(Copyright © 2022, The Author(s), Publisher Elsevier Publishing)

DOI

10.1016/j.multra.2022.100002

PMID

unavailable

Abstract

The 'sharing' business models and on-demand services have been altering city dwellers' travel habits from buying the means of transport to buying mobility services based on needs. The capability to proactively provide personalized services (e.g., travel recommendations and dynamic pricing) is the future of smart multimodal mobility systems, in which the individual mobility prediction technique (predict where/when for a next trip) is a key enabler to achieve that. With the advancement of data collection and computing techniques, the individual mobility prediction problem has been gaining increasing interest, but yet receives little attention in applications and differs in problem definitions and methodologies developed given varied data sources and application contexts. In addition, there are many review studies on collective mobility predictions (e.g., travel demand/traffic flows), but no review study on the individual mobility prediction. To address these issues and fill the gap, the review synthesizes existing studies on individual mobility prediction in transport (data/problem/methodology/applications), identifies remaining research needs, as well as discusses methodological considerations and potential future transport applications. The review highlights the value of individual mobility prediction in driving proactive service provisions and mobility management in multimodal mobility systems.

METHODologically, it is critical to pay more attention to data validation, data specification, and model interpretability in developing learning-based prediction models. Practically, it is of high value to study the individual mobility prediction with arbitrary prediction times (the time when the prediction is made) and prediction horizons (how far in the future).

Language: en

Keywords

Data quality; Deep learning; Individual mobility prediction; Model interpretability; Personalized transportation