Big Trajectory Data Mining: A Survey of Methods, Applications, and Services

Abstract

The increasingly wide usage of smart infrastructure and location-aware terminals has helped increase the availability of trajectory data with rich spatiotemporal information. The development of data mining and analysis methods has allowed researchers to use these trajectory datasets to identify urban reality (e.g., citizens’ collective behavior) in order to solve urban problems in transportation, environment, public security, etc. However, existing studies in this field have been relatively isolated, and an integrated and comprehensive review is lacking the problems that have been tackled, methods that have been tested, and services that have been generated from existing research. In this paper, we first discuss the relationships among the prevailing trajectory mining methods and then, classify the applications of trajectory data into three major groups: social dynamics, traffic dynamics, and operational dynamics. Finally, we briefly discuss the services that can be developed from studies in this field. Practical implications are also delivered for participants in trajectory data mining. With a focus on relevance and association, our review is aimed at inspiring researchers to identify gaps among tested methods and guiding data analysts and planners to select the most suitable methods for specific problems.

A Deep Learning Streaming Methodology for Trajectory Classification

“…Afterwards, a careful selection and analysis of optimal features for the label or labels to be identified followed. Finally, the analysis and selection of several classifiers were performed to achieve high-precision results [5]. Therefore, the overall process of feature selection and classification is not an easy task.…”

“…This can be explained by the fact that neural networks with more hidden layers require more training data; thus, an even larger number of mobility pattern samples was required as an input in these networks. Additionally, we compared our results with other methodologies for trajectory classification, which were illustrated in the recent survey of Wang et al [5]. Well-known classifiers such as Random Forests (RFs) and Support Vector Machines (SVMs) are employed by state-of-the-art methodologies for trajectory classification, on features extracted from the AIS messages or data points of the trajectories.…”

“…On the other hand, trajectory clustering approaches are often employed to form groups of AIS positions with similar spatiotemporal behaviors, uncovering behaviors that are harder to predefine. Although there is an abundance of studies in the literature regarding offline trajectory classification and clustering [1][2][3][4][5], fewer works have focused on steam processing of events in the maritime domain [6][7][8][9][10]. Event processing methodologies are faced with significant challenges when employed on streaming data where the requirements for such applications demand low memory consumption and decreased latencies.…”

Data Harmonization for Heterogeneous Datasets: A Systematic Literature Review

“…In this Systematic Literature Review (SLR), the guidelines were followed from References [28,29]. The research process is divided into three phases.…”

Sequence‐based centrality measures in maritime transportation networks

“…To address the two problems described above, we propose a sequence-based centrality measurement approach to quantify the importance of waypoints in transportation networks. The concept of sequence is frequently used in trajectory analysis, which is a growing research field encompassing topics ranging from travel behaviour discovery and travel pattern generalisation to route detection and network performance evaluation [6]. A sequence is a set of points that form the common path of a group of moving objects [7].…”