Two-step dimensionality reduction of human mobility data: From potential landscapes to spatiotemporal insights

Yunhan Du, Takaaki Aoki, Naoya Fujiwara · 2025

Read Two-step dimensionality reduction of human mobility data: From potential landscapes to spatiotemporal insights by Yunhan Du, Takaaki Aoki, Naoya Fujiwara (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-20929.txt.

Introduction

Tokyo's daily tides of travellers form patterns too dense to read from raw trip records. Yunhan Du, Takaaki Aoki, and Naoya Fujiwara compress those records in two steps. First, a tool from combinatorial geometry turns each hour's origin-to-destination flows into a landscape of peaks and valleys that keeps the imbalance between places. Then a standard statistical squeeze expresses the changing landscapes as a few fixed maps with time-varying weights.

Applied across the pandemic, the method showed overall movement shrinking while the weekday-holiday contrast sharpened. Because space and time separate cleanly, planners can see which districts drive change and when. The authors offer the framework for epidemic control and transport planning wherever cities collect flow data at this scale.

Catalog record

CREATORYunhan Du, Takaaki Aoki, Naoya Fujiwara
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-20929

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Two-step dimensionality reduction of human mobility data: From potential landscapes to spatiotemporal insights Yunhan Du, Takaaki Aoki, Naoya Fujiwara Abstract Understanding the spatiotemporal patterns of human mobility is crucial for addressing societal challenges, such as epidemic control and urban transportation optimization. Despite advancements in data collection, the complexity and scale of mobility data continue to pose significant analytical challenges. Existing methods often result in losing location-specific details and fail to fully capture the intricacies of human movement. This study proposes a two-step dimensionality reduction framework to overcome existing limitations. First, we construct a potential landscape of human flow from origin-destination (OD) matrices using combinatorial Hodge theory, preserving essential spatial and structural information while enabling an intuitive visualization of flow patterns. Second, we apply principal component analysis (PCA) to the potential landscape, systematically identifying major spatiotemporal patterns. By implementing this two-step reduction method, we reveal significant shifts during a pandemic, characterized by an overall declines in mobility and stark contrasts between weekdays and holidays. These findings underscore the effectiveness of our framework in uncovering complex mobility patterns and provide valuable insights into urban planning and public health interventions. 0 0 footnotetext: Google scholar profiles: Yunhan Du https://scholar.google.com/citations?user=zgv6kzkAAAAJ&hl=en ; Takaaki Aoki https://scholar.google.com/citations?user=e8x3baAAAAAJ&hl=en ; Naoya Fujiwara https://scholar.google.co.jp/citations?user=qmt1jPMAAAAJ&hl=en keywords Human mobility, potential landscape, principal component analysis, mobility patterns, COVID-19 pandemic 1 Introduction Humans are inherently mobile beings who frequently travel from one location to another for various daily activities, such as commuting, shopping, entertainment, and education. These individual movements collectively shape the intricate dynamics of urban mobility. Spatiotemporal patterns of human flow are crucial for a wide range of applications, including controlling the spread of epidemics Jia et al. 2020 ; Viboud et al. 2006 ; Hazarie et al. 2021 , understanding social and economic issues such as social segregation Amini et al. 2014 ; Moro et al. 2021 ; Yang et al. 2023 and urban gender disparities Gauvin et al. 2020 ; Collins et al. 2023 , and integrating public transportation systems with urban design plans to enhance mobility efficiency Vazifeh et al. 2018 ; Xu et al. 2021 ; Bongiorno et al. 2021 ; Gallotti et al. 2016 . Human flow is not static but fluctuates dynamically over time. Therefore, understanding the evolution of these patterns is crucial. For example, on weekdays, cities exhibit distinct human flow patterns during morning and evening rush hours, whereas on weekends, movement trends differ significantly. Furthermore, external factors such as natural disasters and pandemics can drastically alter these patterns. Natural disasters, such as earthquakes or floods, can abruptly disrupt natural human movement, triggering mass evacuations or substantial changes in commuting behaviors. Similarly, the outbreak of a pandemic, such as Coronavirus Disease 2019 (COVID-19) Zhu et al. 2020 , can cause unprecedented shifts in human mobility, with lockdowns and travel restrictions drastically reducing movement and reshaping human flow patterns on a global scale. These temporal variations in human flow result from a complex interplay of factors, including economic activity, social events, environmental conditions, and policy interventions. Examining these patterns over time provides valuable insights into the underlying mechanisms driving human mobility. Recent advancements in data collection and processing technologies have enabled the analysis of large-scale mobility data at high spatiotemporal resolutions, facilitating the exploration of dynamic mobility patterns and the identification of underlying structures in human flow. We live in an era where vast amounts of human mobility data are generated from various sources Blondel et al. 2015 ; Zheng 2015 ; Zhao et al. 2016 ; Zhu et al. 2018 ; Yabe et al. 2022 . Traditionally, mobility data have been gathered through censuses and surveys that collect information on where people live and work. With the increasing prevalence of mobile phones, call detail records (CDRs) have become an important data source. These records, provided by mobile network operators, track the location of a signal tower when a phone call or text message is sent. Additionally, global positioning system (GPS) data collected from GPS-enabled devices and smartphones offer increasingly detailed location information. Social media platforms also contribute valuable mobility data through user interactions and geotagged posts. Despite the availability of vast amounts of data, understanding the spatiotemporal patterns of human flow remains challenging owning to the large volume and complexity of such data. In previous research, mobility data have often been reduced down to temporal trends of macroscopic statistics, such as displacement and waiting time distributions Brockmann et al. 2006 , radius of gyration González et al. 2008 , number of visited places and visitation frequency Song et al. 2010a , predictability Song et al. 2010b , motifs Schneider et al. 2013 , and mobility synchronization Santana et al. 2023 . Some studies have examined the dynamics of one or more of these metrics Kraemer et al. 2020 ; Hasan et al. 2013 ; Lu et al. 2012 ; Geng and Yang 2017 ; Wang and Taylor 2016 ; Heiler et al. 2020 ; Kalleitner et al. 2022 ; Santana et al. 2023 , revealing common patterns in human mobility. However, these data processing approaches often mitigate information overload by reducing detail, resulting in the loss of location-specific information. In other instances, data have been transformed into spatial distributions of temporal populations Bengtsson et al. 2011 ; Li et al. 2016 ; Lloyd et al. 2017 , preserving location-specific details but losing crucial mobility information, such as trip distances and their origins. Additionally, data have been used to fit general models of human mobility, including the gravity model Zipf 1946 , the intervening opportunities model Stouffer 1940 ; Fik and Mulligan 1990 ; Akwawua and Pooler 2001 ; Noulas et al. 2012 ; Sim et al. 2015 , the radiation model Simini et al. 2012 ; Liang et al. 2013 ; Yan et al. 2014 ; Ren et al. 2014 ; Kang et al. 2015 ; Varga et al. 2018 ; Liu and Yan 2019 ; Liu and Yan 2020 ; Du and Fujiwara 2022 , and the exploration and preferential return (EPR) model Song et al. 2010a ; Pappalardo et al. 2015 ; Schläpfer et al. 2021 ; Barbosa et al. 2015 ; Alessandretti et al. 2018 ; Lv et al. 2021 . Although these models offer valuable insights, systematic examination of the dynamics of human mobility patterns across different contexts is limited. In summary, previous approaches have failed to retain the full complexity of human mobility data. Methods that aggregate data into macroscopic statistics often lose spatial specificity, whereas those emphasizing spatial distributions omit essential trip information, such as origins and distances. Furthermore, model-based approaches abstract real-world patterns to fit theoretical constructs, which may not fully capture the intricacies of human movement. To address those existing issues, this study proposes a two-step reduction method for human mobility data, enabling the effective visualization of spatiotemporal patterns in human flow (Fig. 1 ). As shown in the middle panel of Fig. 1 , the first step involves utilizing the concept of the potential landscape of human flow Aoki et al. 2022 ; Aoki et al. 2024 ; Mazzoli et al. 2019 ; Shida et al. 2022 ; Yang et al. 2024 . In our case, the potential landscape is generated through applying the combinatorial Hodge theory

Read the complete text