Naslov
Identification Of Features For Trajectory Segmentation According To The Transport Mode

Autori
Erdelić, Martina ; Erdelić, Tomislav ; Carić, Tonči ; Mardešić, Nikola

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, prošireni sažetak, znanstveni

Izvornik
TODO International Conference on Open Data proceedings / - , 2022, 1-5

Skup
TODO International Conference on Open Data

Mjesto i datum
Zagreb, Hrvatska, 28-2.12.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
urban mobility ; transport mode ; smartphone sensor data ; trajectory segmentation ; feature selection

Sažetak
A transport network is a complex system whose analysis requires data from transport network users. The influence of the growing degree of urbanization and population places transport as one of the main factors affecting the quality of life in large cities. Therefore, research fields such as urban mobility, energy consumption, pollution reduction and security are often the subject of scientific research. All these research fields are directly or indirectly related to the mobility of transport network users. An in-depth trajectory analysis, which can be supported by clustering or classification methods, can describe the mobility of users through the transport network. User trajectories often contain several connected transport modes that users use, moving from the origin to the destination. Hence, such trajectories need to be segmented before further processing. Therefore, there is a need to develop a trajectory segmentation method that will recognize the mode transfer point. In this paper, several features and their influence on the accuracy of mode transfer point detection were tested. The time and frequency domain of trajectory data are used for feature extraction. Features are selected using the ANOVA F value, and different feature sets are tested using the transition state matrices approach. A large-scale dataset of smartphone sensor data recorded through seven months in 2017 is used for that purpose. Based on the results of the trajectory segmentation method using the identified set of features, it can be concluded that the computed and selected features can describe the user behavior pattern when changing the transport mode.

Izvorni jezik
Engleski

Znanstvena područja
Tehnologija prometa i transport, Interdisciplinarne tehničke znanosti

POVEZANOST RADA