SOCIAL MEDIA INFRASTRUCTURE AND URBAN BOUNDARIES
// rethinking the administrative boundaries of the pearl river delta in china
Effective governance has always relied on fixed borders in order to facilitate the core functions of defense and taxation. For the purposes of government administration, the boundaries of cities and regions are static lines presented on a map. In the current globalized era, however, this conception of urban or regional boundaries is problematic. The economic, social, and cultural pull of cities today is such that individuals regularly shuttle back and forth between cities, counties, and provinces in the course of their work week.
Large datasets, such as those generated by social media activity, can can be used to illustrate the dynamic and elastic nature of urban or regional boundaries. Using user-generated social media data as the foundation, this project is an investigation of urban boundaries and social media infrastructure in the Pearl River Delta (PRD) of China. the departure point of this project is that the 30 counties that comprise the administrative boundaries of the PRD do not reflect the movement of the 60 million inhabitants of this dense and interconnected region. Data from the Chinese micro-blogging service Weibo provides a means to understand the interplay between the fixed borders and dynamic user-generated boundaries.
// the built environment and border crossings
In order to understand the daily movement of people throughout the PRD, we mapped the origin and destination locations of border crossing that occurred within a single day (24 hr period) within the PRD. The initial visualizations illustrate that all of the border crossing check-ins conform to the built environment of the PRD to a high degree. The built landscape of the PRD is interconnected, creating a network of cities and development around the inner ring of the region. Similarly, all of the border crossing check-ins reflect this interconnected ring of development.
//hotspot analysis of border crossings
Refining the check-in data further, the highest density areas are shown to conform to major urban areas such as Shenzhen, Macao, Hong Kong, and Guangzhou. While perhaps not surprising, these nascent clusters are a starting point for visualizing the new social boundaries of the PRD region. By separating the check-in data by time of day, new boundaries of home and work emerge. Home boundaries refer to those check-ins that occur from 10pm to 5am and work boundaries to the hours of 9am to 6pm.
//cluster analysis of major city hotspots
The following maps are based on the identified home and work hotspot clusters for three cities: Shenzhen, Guangzhou, and Dongguan. These clusters are depicted on the maps in yellow for home and teal for work. The maps depict the network of locations where people cross boundaries to get to these cluster locations. A hotspot analysis of the crossings to and from these home and work clusters illustrates the locations in which the people most frequently travel between the identified home and work clusters. For this hotspot cluster analysis, the red areas highlight the most significant density of check-ins to and from the cluster regions, while the grey areas are less significant clustering of check-ins based on density.
//methodology
This was a collaborative group project for a visual studies course at the Graduate School of Architecture, Planning and Preservation at Columbia University; other group members included: Yuheng Cai, Houman Saberi, and Junda Chen.
A database of weibo check-ins throughout the PRD, provided for the class, served as the starting point for this research. This consisted of two separate databases that we then combined using python to get a dataset that linked user ID to coordinate information, poiid (unique location ID), and time of check-in information. In order to study the boundary crossing we needed to trace each user's check-in information to see whether or not he/she has crossed the boundary. We separated the database by each user ID and arranged it by check-in time. Using the Pearl River Delta (PRD) boundaries on GIS, we matched each unique poiid number with a county number within the PRD. We then ran a loop on python to compare each of the adjacent check-ins' county number. If two adjacent check-ins have different county number, we then ran an additional test to see if this boundary crossing occurred within one day.
A coupled check-in contains two check-ins. The first check-in is named as starting point and the second check-in is named as ending point. We first load this boundary crossing check-in data directly to GIS and created boundary crossing points and lines. But there was too much data showm in the map and it was hard to distinguish any pattern. A hotspot analysis of all of the check-in locations that crossed boundaries was used to better understand and analyze the data set. The ArcGIS hotspot statistical analysis tool was used to find hotspots of starting and ending locations for border crossings. However, in order to further wanted to narrow down our scope and understand travel and commute patterns of those who move throughout the PRD region, we used excel to filter the coupled check-in dataset by time of day. All check-ins whose starting points are within 9:00am – 6:00pm and ending points within 10:00pm – 5:00am were selected to a new data set. An additional hotspot analysis was then done using this new data set.
We found from the home and work hotspot map that there was a very clear boundary between each of the home hotspots and work hotspots. We then sought out to understand the travel patterns between selected work and home hotspots with other hotspots and major cities. We selected five hotspots from three cities: Shenzhen, Guangzhou, and Dongguan. Shenzhen and Guangzhou are both big cities that contain multiple work and home hotspots. But in Dongguan, a small city between Shenzhen and Guangzhou, there was only one home hotspot and no work hotspot.
download the full group report here