Street Network Analysis
- Introduced by
- Geoff Boeing (OSMnx); graph-theoretic street analysis tradition
- Year
- 2017
- Type
- Graph-theoretic measurement of street-network structure and connectivity
- Family
- Spatial network / street-pattern metrics
What does it do?
Street network analysis treats a city's streets as a mathematical graph — intersections as nodes, street segments as edges — and measures its structure with graph-theoretic indicators of connectivity, density, centrality, and efficiency.
From this representation come the metrics that distinguish a permeable grid from a tree-like cul-de-sac suburb: intersection density, average node degree, the share of dead-ends, betweenness centrality, and circuity (how much longer network routes are than straight lines). Tools such as Geoff Boeing's OSMnx made it routine to download, model, and analyse the street network of any place on Earth from OpenStreetMap, turning street-pattern analysis into a reproducible, comparative science of urban form.
The intuition behind the method
How streets connect shapes almost everything about how a place works — how far you must walk or drive between two points, whether traffic concentrates on a few arteries, how easily pedestrians can find a direct route. Picture the difference between a dense gridiron, where you can turn at every block and most paths are short and direct, and a dendritic suburb of looping collectors feeding cul-de-sacs, where even neighbours are a long detour apart. Street network analysis makes those differences measurable: it counts how many streets meet at the average intersection, how dense intersections are per square kilometre, which streets carry the most through-routes, and how circuitous travel is. Because the street network is just a graph, the whole toolkit of graph theory becomes available to describe and compare urban form.
History
Graph-theoretic description of transport networks dates to the quantitative-geography work of Garrison, Kansky, and Haggett in the 1960s, who introduced connectivity indices such as alpha, beta, and gamma. Space syntax, developed by Bill Hillier and colleagues from the 1980s, added configurational measures linking street structure to movement, while the rise of OpenStreetMap and open spatial software made large-scale, reproducible analysis feasible. Geoff Boeing's 2017 OSMnx package was a turning point: it let researchers download, construct, correct, and analyse the street network of any place from OpenStreetMap with a few lines of Python, computing connectivity, centrality, and circuity metrics at scale and enabling worldwide comparative studies of street patterns and sprawl.
When to use it
Use street network analysis when the structure of streets themselves matters — comparing the connectivity and walkability of neighbourhoods or whole cities, feeding network metrics into sprawl and walkability indices, identifying high-traffic corridors, or studying how street patterns relate to travel behaviour, safety, and emissions. It is the right tool when you have or can obtain a clean street graph and want reproducible, comparable metrics across places. It is less appropriate when the question is really about routing individual trips with realistic costs (use a routing engine), when OpenStreetMap coverage is poor or inconsistent in the study area, or when configurational human-movement prediction is the goal, where space syntax's axial and angular analysis may be more suitable.
Limitations and considerations
- Results depend on OpenStreetMap completeness and consistency, which varies widely between regions.
- Graph simplification choices (how intersections and multi-edges are handled) affect the computed metrics.
- Topological metrics ignore street width, sidewalk quality, traffic, and the lived pedestrian experience.
- Betweenness and other global metrics are sensitive to the chosen study-area boundary and edge effects.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Brings the full power of graph theory to bear on urban form with well-defined, comparable metrics.
- Reproducible and global: OSMnx lets anyone model and analyse any city's network from OpenStreetMap.
- Connectivity, centrality, and circuity metrics directly inform sprawl, walkability, and transport analyses.
- Scales from a single neighbourhood to thousands of cities for large comparative studies.
Applications
- Comparing the connectivity and grid fineness of neighbourhoods, cities, and entire countries.
- Supplying intersection density and connectivity inputs to sprawl and walkability indices.
- Identifying high-betweenness corridors for traffic, transit, and retail-location analysis.
- Measuring circuity to assess travel efficiency, walkability, and car dependence.
- Studying relationships between street-network form and road safety, emissions, and public health.
How it works
How it works
The street system is modelled as a graph G = (V, E) in which nodes V are intersections and dead-ends and edges E are the street segments connecting them, annotated with length, geometry, and often travel speed or direction. Building this graph correctly — simplifying interstitial nodes, handling one-way streets and multi-edges, and projecting to an equal-distance coordinate system — is the foundational step, and OSMnx automates it from raw OpenStreetMap data for any city or boundary.
Formula
G = (V, E),\quad V=\text{intersections},\ E=\text{street segments}Structural metrics summarize how well connected and fine-grained the network is. Average node degree k = 2|E|/|V| and the share of three- and four-way intersections describe how many streets meet at each junction; intersection density ρ = |V|/A measures grid fineness per unit area; and connectivity indices such as the gamma index γ = |E|/[3(|V|−2)] express edges as a fraction of the maximum possible. Together with the proportion of dead-ends, these distinguish permeable grids (high degree, high density, few cul-de-sacs) from sprawling, tree-like patterns.
Formula
k = \frac{2|E|}{|V|},\qquad \rho_{int} = \frac{|V|}{A},\qquad \gamma = \frac{|E|}{3(|V|-2)}Centrality measures identify which streets and intersections are structurally important. Betweenness centrality C_B(v) counts the share of shortest paths between all node pairs that pass through a node (or edge), highlighting the through-routes and bottlenecks that carry disproportionate movement. High-betweenness segments tend to attract traffic and commerce, so the measure links network configuration to flow, congestion, and the location of activity — a quantitative cousin of the integration ideas in space syntax.
Formula
C_B(v) = \sum_{s\ne v\ne t} \frac{\sigma_{st}(v)}{\sigma_{st}}Circuity (detour ratio) ζ is the total network distance between locations divided by the total straight-line distance, ζ = Σ d_net / Σ d_euc. A value near one means the network lets people travel almost as the crow flies — efficient and walkable — while higher values reveal indirect, detour-heavy patterns typical of cul-de-sac sprawl. Because it directly measures travel inefficiency, circuity is a compact indicator of how a street pattern translates into longer trips and greater car dependence.
Formula
\zeta = \frac{\sum_{ij} d^{net}_{ij}}{\sum_{ij} d^{euc}_{ij}}Common pitfalls
- Computing metrics on an unsimplified or unprojected graph, inflating node counts and distorting density and degree.
- Comparing intersection density or betweenness across areas clipped with different boundaries without correcting edge effects.
- Trusting metrics in regions where OpenStreetMap street coverage is sparse or inconsistent.
- Equating high topological connectivity with good walkability while ignoring traffic, sidewalks, and crossings.
Frequently asked questions
How is street network analysis different from space syntax?
Both analyse street structure as a network, but they model it differently and ask different questions. Street network analysis here uses the primal graph — intersections as nodes, segments as edges — and computes metrics like intersection density, node degree, circuity, and betweenness, emphasizing sprawl and connectivity. Space syntax uses a dual, configurational representation (axial lines or segment angular analysis) and measures integration and choice to predict human movement. They overlap in spirit but differ in graph construction and theoretical focus.
What does circuity tell me that intersection density does not?
Intersection density measures how fine-grained the grid is — how many junctions per unit area — but a network can be dense yet still force detours. Circuity measures the actual travel penalty: the ratio of network distance to straight-line distance between locations. A grid has circuity near one (you can almost go as the crow flies), while a dense but loop-and-cul-de-sac layout can still have high circuity. Together they describe both how fine and how efficient the network is.
Why is OpenStreetMap data quality such a concern?
Street network metrics are only as good as the underlying graph, and OpenStreetMap is crowd-sourced, so its completeness and accuracy vary by region. In well-mapped cities the network is highly reliable; in under-mapped areas missing streets or misclassified paths can distort node counts, density, and connectivity. Analysts should check coverage, use consistent graph-simplification settings, and be cautious when comparing places that may be mapped to different standards.
Related methods
Sources
Boeing, G. (2017). OSMnx: New methods for acquiring, constructing, analyzing, and visualizing complex street networks. Computers, Environment and Urban Systems, 65, 126–139.
Open DOI referenceCopied.