Shanghai model maps how streets could widen access to daily services

A street-by-street assessment of 1,396 residential communities in Shanghai’s Huangpu District has identified where daily services are easy to reach, where they are not, and which urban changes could close the gaps. The research combines public-service locations, community boundaries, census information and OpenStreetMap road networks to measure access as a “15-minute community life circle,” a planning idea in which residents should be able to reach essential services through a short journey from home. Its findings suggest that the physical layout of streets matters more than development density alone in determining who can obtain convenient access.

The study, published in Humanities and Social Sciences Communications, divides communities into three service categories using operational road-network distances of 500 and 800 metres. Category A represents the highest service level, Category B an intermediate level and Category C the lowest. Rather than assuming that every point within a straight-line radius has the same practical access, the method calculates shortest paths along the street network. This distinction is important in a compact city where barriers such as major roads, complex blocks or missing connections can make a nominally short trip difficult.

The results reveal pronounced spatial variation across Huangpu. Communities in Category A contain 54.60% of the district’s population and are concentrated in the core commercial district and along the waterfront. Category C communities account for 11.32% of residents and tend to occur in older residential areas in the northwest. The pattern shows how a district can possess extensive service infrastructure and still contain neighborhoods in which access remains comparatively poor. Distribution is therefore not simply a question of how many facilities exist, but also of where residents live and how efficiently the street system connects them to those facilities.

To determine which features best explain the observed differences, the researchers trained a C5.0 decision-tree model using the classified communities. The model achieved 81.74% accuracy on its test set. Road-network density emerged as its strongest predictor, with an importance value of 0.42. Taken together, transportation-related indicators accounted for 0.53 of the model’s predictive importance, more than twice the 0.19 attributed to indicators of development intensity. In practical terms, the analysis points to connectivity as the strongest lever for improving access rather than simply adding more development or treating density as the principal obstacle.

The decision tree converted those relationships into classification rules that can help planners identify communities meeting the model’s conditions. Such a tool does not determine the policy response on its own, but it can reveal where the existing combination of services and transport is likely to fall below a target. For a district such as Huangpu, this creates a way to move from a general statement that access is uneven to a community-level diagnosis. The model also provides a consistent basis for comparing possible interventions, provided planners retain the assumptions and spatial conditions used in the analysis.

Using the diagnosed network as a baseline, the authors tested two predefined policy scenarios. The first assumed transportation improvement, while the second adjusted development intensity. Both scenarios shifted population toward the higher service categories in the model, but the transport intervention produced the larger predicted effect. Its simulations increased the share of residents in Category A by 7.98 percentage points and reduced the Category C share by 4.88 points. Under the development-intensity scenario, the Category A share rose by 5.24 percentage points, while the Category C share fell by 4.16 points.

The contrasting results sharpen the policy question. More intense development is associated with changing service levels, but the study’s predictor analysis indicates that street-network conditions carry greater weight in the model. Improving connectivity can therefore be expected to alter access more substantially in the simulated system, although the authors emphasize that the figures are model predictions tied to the specified parameters and spatial conditions. The scenarios should not be read as guaranteed outcomes or as evidence that one intervention will always outperform another in every high-density district.

Even so, the work offers a transferable method for a global planning challenge. Cities pursuing 15-minute neighborhoods often face pressure to place schools, healthcare facilities, shops, parks and other amenities within convenient reach of residents. A simple radius drawn on a map can overstate accessibility because it ignores the routes people actually travel and the physical constraints of the built environment. Network-based measurement instead evaluates each community through the connections available on the ground, making the diagnosis more closely aligned with everyday mobility.

The study also highlights the need for careful interpretation. The analysis uses secondary spatial and aggregate census data and did not involve human participants, while the decision tree and scenario results depend on the selected service categories, distance thresholds, model parameters and observed conditions in Huangpu. It identifies communities that may warrant priority review and compares the two predefined scenarios, but it does not capture every aspect of lived access, including personal mobility, service quality, affordability or barriers faced by particular residents. The authors describe the decision-tree rules and scenario predictions as supplementary information for planning review rather than as a complete decision system.

For Huangpu, the immediate message is that accessibility is a distributional problem embedded in the urban fabric. The most advantaged communities cluster around the commercial core and waterfront, while lower-service populations are concentrated in older northwestern neighborhoods. A network model that links facilities, homes and streets turns those contrasts into testable policy questions: where connections are weak, what specific barriers impede travel, and which changes would bring more residents into the highest service tier? By showing that transport indicators outweigh development-intensity indicators in the model, the research points toward connectivity as a promising place for planners to begin. The next step would be to use the diagnostic output to design and evaluate real interventions, rather than treating the simulated gains as an already achieved urban outcome.

Subject of Research: Measuring and simulating 15-minute community service access in Shanghai’s Huangpu District

Article Title: Diagnosing 15-minute community life circle service levels and simulating policy scenarios in a high-density urban core: Shanghai’s Huangpu District

Article References: Lei, Y., Zou, T., Tian, J., An, B., & Shen, F. (2026). Diagnosing 15-minute community life circle service levels and simulating policy scenarios in a high-density urban core: Shanghai’s Huangpu District. Humanities and Social Sciences Communications. https://doi.org/10.1057/s41599-026-09365-4

Image Credits: AI Generated

DOI: 10.1057/s41599-026-09365-4

Keywords: 15-minute cities, urban accessibility, Huangpu District, Shanghai, public services, transport networks, decision trees, urban planning, community life circles, policy simulation, Diagnosing, minute