The increasing availability of longitudinal individual human mobility traces enables the disaggregate analysis of temporal properties of mobility patterns. The objective of this study is to identify distinctive market segments in terms of habitual temporal travel patterns of public transport users. First, travel patterns are clustered using a Kmeans approach followed by grouping the resulting patterns into a small number of profiles using a hierarchical clustering method. Second, we construct user-week vectors that are then clustered using a Gaussian Mixture Model approach. We apply our clustering analysis to the multi-modal public transport system of Stockholm County, Sweden, using data from more than 3 million smart card-holders. Our clustering analysis resulted in 10 day-of-the-week patterns with their composition varying across the county. In addition, we identify the following hour-by-hour weekly profiles:'Weekly commuters', 'Lower peaks','Late travellers', 'Early birds' and 'Flat curve'. The behavior represented by 'Weekday commuters' and 'Lower peaks' is most persistent over weeks. We demonstrate how a better understanding of user travel patterns offers policy makers, service planners and providers with enhanced opportunities to understand and cater for diverse market segments, for example by means of tailored fare products.

The analysis of longitudinal travel data enables investigating how mobility patterns vary across the population and identify the spatial properties thereof. The objective of this study is to identify the extent to which users explore different parts of the network as well as identify distinctive user groups in terms of the spatial extent of their mobility patterns. To this end, we propose two means for representing spatial mobility profiles and clustering travellers accordingly. We represent users patterns in terms of zonal visiting frequency profiles and grid-cells spatial extent heatmaps. We apply the proposed analysis to a large-scale multi-modal mobility dataset from the public transport system in Stockholm, Sweden. We unravel three clusters - Locals, Commuters and Explorers - that best describe the zonal visiting frequency and show that their composition varies considerably across users’ place of residence. We also identify 15 clusters of visiting spatial extent based on the intensity and direction in which they are oriented. A cross-analysis of the results of the two clustering methods reveals that user segmentation based on exploration patterns and spatial extent are largely independent, indicating that the two different clustering approaches provide fundamentally different insights into the underlying spatial properties of individuals’ mobility patterns. The approach proposed and demonstrated in this study could be applied for any longitudinal individual travel demand data.

Urban and regional areas worldwide exhibit a complex and uneven distribution of activities with certain areas attracting more people during different time periods. In this study we systemically classify different parts of the urban area which are most attractive as measured by their ability to attract visitors. A weekly visiting profile is constructed for each travel demand zone and thereafter clustered to identify areas with common attraction patterns. We leverage on the availability of longitudinal individual mobility traces in the form of smart card data transactions. We apply our method to the case study of the multi-modal public trans-port system of the Stockholm urban agglomeration area. The results of our clustering based on the weekly visiting profiles reveal four distinctive types of visiting attraction based on the intensity and temporal distribution of activities performed. The results of this study can be used to inform planners and decision makers about the main activity locations of travellers and how their temporal patterns vary across the metropolitan area and the design of related policies.

Background

This project followed-up on a project called FairAccess which was granted in Trafik och Region 2018.In FairAccess, we processed Access card data and performed a sequence of inferences to derive timedependent origin-destination matrices for the entire Region Stockholm system. Tap-in records werematched with corresponding inferred tap-out locations and time stamps for about 80% of all records.Moreover, we implemented an algorithm to generate a journey database based on our transferinference method. We used the outputs of this process to evaluate the impacts of the fare schemechange (i.e. from zone-based to flat fare) on different user profiles. Access card products and zonalattributes were used for analysing policy impacts on different market segments.The “Unravelling Mobility Patterns using Longitudinal Smart Card Data” project was granted on May27, 2020 and the contract was signed on July 17, 2020. In this project, we capitalise on the capabilitiesof the inferences performed in previous work to conduct a series of market segmentation andadvanced data analytics to empirically analysis demand patterns for public transport in the StockholmCounty. The growing travel demand in Stockholm County is accompanied by an increased diversity ofsub-centres within the region as well as in individual travel patterns. It is thus increasingly importantto understand how demand patterns evolve over time, what the key market segments are and howdifferent users are affected by changes in service provision. The latter is studied in the contact of theopening of the Citybanan project.As stated in the SLL Research and Innovation Plan, the development of transport solutions for theStockholm region requires new knowledge regarding travellers’ needs and preferences, and theimpacts for different types of travellers.