TL;DR
This paper introduces a novel framework to analyze crowd activity dynamics in commercial streets by combining trajectory data with format information. By identifying stay points and trajectory clusters, the study reveals how different commercial formats influence crowd behavior, offering valuable insights for urban planning and sustainable development.
What Problem Does This Paper Address?
Commercial streets are vital to urban economies, but their vitality is threatened by changing consumer habits and the rise of online retail. Understanding how people interact with these spaces is crucial for their survival and development. While previous research has looked at crowd movement and its relationship with the built environment, there's a gap in understanding the nuanced interplay between specific commercial formats and the dynamic spatial and temporal behaviors of crowds. This paper aims to bridge this gap by analyzing crowd activities not just as points on a map, but as dynamic journeys influenced by the types of businesses present.
Methods
The authors propose a three-step framework:
- Stay Point Identification: Using the ST-DBSCAN algorithm, they identify where people "stay" within the commercial street, defining a stay as movement within 10 meters for 5 minutes. This helps understand where people pause and engage with the environment.
- Road Network Matching: To simplify complex trajectories and improve accuracy, they employ a multi-criteria dynamic programming map-matching (MDP-MM) algorithm. This process converts raw GPS data into cleaner, road-aligned movement paths.
- Trajectory Clustering: A Convolutional Autoencoder (CAE) is used to cluster crowd trajectories, transforming them into "images" for analysis. This allows for the identification of distinct patterns of movement and activity, revealing four main types of crowd trajectories.
Finally, a Partial Least Squares Regression (PLSR) model is used to quantify the relationship between crowd activity characteristics (number of tourists, per capita stays) and commercial street format variables (spatial distance, quantity, and diversity of formats).
Datasets
The study utilizes data from the Lao Men Dong Commercial Street in Nanjing, China.
- Format Data: Collected from the Gaode Developer platform, including format names, categories, and spatial locations.
- Road Network Data: Also from the Gaode Developer platform, detailing road lengths, widths, and spatial locations.
- Crowd Trajectory Point Data: Collected using Wi-Fi probe devices, capturing time-space position information. Data was collected from 13:00 to 20:00 on weekdays and weekends between November 26 and December 7, 2022, to avoid holidays and extreme weather.
Key Insights and Ideas
- Weekends vs. Weekdays: Crowd stay activity is significantly higher on weekends, with people more willing to linger in catering and leisure spaces. Weekday activities are more scattered, focusing on retail and cultural formats.
- Centrality Matters: The central axis of the commercial street exhibits the highest crowd stay intensity, primarily due to the concentration of catering and leisure formats. Outer areas with fewer format types see lower dwell times.
- Trajectory Patterns Reveal Preferences: Four distinct crowd trajectory clusters were identified, highlighting different movement and stay behaviors. Cluster 1, characterized by strong spatial mobility and short stays in retail/recreational facilities, represents the dominant activity pattern.
- Format Influence is Nuanced: While the mix and spatial location of formats have a relatively weak or even negative impact on overall crowd activity, the intensity of specific formats (food, shopping, tourist attractions, public services) has a significant positive effect, drawing people in for longer stays.
- Entrance/Exit Dynamics: People tend to enter and leave Lao Men Dong Commercial Street from the north entrance, likely due to its prominence as a landscape symbol, despite other designated entrances.
Key Results
- Stay Points: Over 76,000 stay points were identified, revealing temporal and spatial patterns of crowd presence.
- Trajectory Clusters: Four distinct crowd trajectory clusters were identified, with Cluster 1 accounting for 40.49% of active people.
- Format Impact: Food, shopping, tourist attraction, and public service formats showed a significant positive effect on crowd activities. Leisure formats also had a positive impact on stay activities, while the mix of formats and spatial distance had a negative effect.
- Weekend Intensity: Weekend stay activity intensity was much higher than on weekdays.
Limitations and Future Work
The authors acknowledge several limitations:
- Scope of Analysis: The study focused on formats and ignored other environmental factors like transparency and green viewing rates.
- Data Collection Period: Data was collected only during peak hours (13:00-20:00) in winter, limiting insights into 24-hour activity patterns and seasonal variations. Future work should collect data across different seasons and times.
- Individual Tourist Preferences: The study primarily analyzed common trajectory patterns, potentially overlooking the unique needs and preferences of individual tourists, which could influence format choices. Future research should consider these special trajectories.
- Data Handling: Occasional tourist groups guided by tour guides were observed but could not be fully processed due to Wi-Fi probe data limitations. Future work could explore methods to better handle such data.
The authors suggest that their framework is applicable to various spatial scales beyond commercial streets, offering a promising avenue for future research in urban planning and sustainable development.
Summary updated