Dalarna University's logo and link to the university's website

Transportation research has benefited from GPS tracking devices since a higher volume of data can be acquired. Trip information such as travel speed, time, and most visited locations can be easily extracted from raw GPS tracking data. However, transportation modes cannot be extracted directly and require more complex analytical processes. Common approaches for detecting travel modes heavily depend on manual labelling of trajectories with accurate trip information, which is inefficient in many aspects. This paper proposes a method of semi-supervised machine learning by using minimal labelled data. The method can accept GPS trajectory with adjustable length and extract latent information with long short-term memory (LSTM) Autoencoder. The method adopts a deep neural network architecture with three hidden layers to map the latent information to detect transportation mode. The proposed method is assessed by applying it to the case study where an accuracy of 93.94% can be achieved, which significantly outperforms similar studies.

The widespread adoption of Global Positioning Systems (GPS) in transportation has significantly contributed to the understanding of human behaviour, enabling the extraction of crucial travel information. However, the exploration of transportation modes using GPS data remains an under-researched domain due to its intricate analytical demands. While various methods, ranging from rule-based approaches to advanced machine learning algorithms, have been employed to identify transportation modes from GPS data, most have been tested on limited labelled datasets. This study introduces an innovative clustering method that combines multi-criteria decision-making, network analysis, and the meta-heuristic algorithm of particle swarm optimization to effectively cluster transportation modes. Pioneering a hybrid approach, the study utilizes elements from the Analytic Network Process (ANP) super matrix in conjunction with transportation modes as variables, harnessing the particle swarm optimization (PSO) algorithm with a fully unlabelled dataset. The compelling findings underscore the model's effectiveness, achieving an impressive accuracy rate exceeding 88% in transportation mode clustering.

Human mobility behaviour is far from random and can be predictable. Predicting human mobility behaviour has the potential to improve location selection for facilities, transportation services, urban planning, and can be beneficial in providing more efficient sustainable urban development strategies. However, it is difficult to model urban mobility patterns since incentives for mobility is complex, and influenced by several factors, such as dynamic population, weather conditions. Thus, this paper proposes a prediction-oriented algorithm under the framework of a Hidden Markov Model to predict next-location and time-of-arrival of human mobility. A comprehensive evaluation of these two schemes for the representation of latent and observable variables is discussed. In conclusion, the paper provides a valuable contribution to the field of mobility behaviour prediction by proposing a novel algorithm. The evaluation shows that the proposed algorithm is stable and consistent in predicting the next location of users based on their past trajectories. © 2024 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.