Showing all amenities

Exploring Urban Micro-Agglomeration Patterns

Jane Jacobs, extending on ideas proposed by Warren Weaver, suggested that cities are problems of organized complexity. This view of the city embraces organic agglomeration, diversity, and proximity, and presents the challenge of interpreting and assessing this complexity, especially on a local-scale.

placeSpace provides a lens to explore the complexity of the urban environment by spatial distribution and agglomeration of amenities at a local urban scale. Through this interactive data visualization tool, users can gain a better understanding of how amenities agglomerate, co-locate and define areas within the city.

The platform currently only supports London, but can be extended to analyze other cities.

Amenity Distribution

The presence of different amenities has become an important aspect of livable and walkable cities. Their location and distribution can tell about the environmental, social, and economic aspects of a particular place.

On the graph bellow you can visualize the how the total number of amenities in the city compare with each other. On the right, a heat map is shown representing the concentration of amenities.

  • Hover over an element to see its total count.
  • Click on an element to see its distribution over London.
  • To Control level of detail, click on buttons on bottom of graph.

Defining Areas

To identify local scale agglomeration patterns different methods can be used, each yielding different results based on how each method identifies a spatial cluster. Here two density clustering methods are explored: DBSCAN and HDBSCAN. Click on the buttons bellow to see how they differ.

HDBSCAN produces more homogeneous clusters, and their boundaries seem to correspond well to known areas in the city. For the rest of the analysis the results from this clustering method are used.

Total Number of Areas Identified :

Co-location Network

After identifying areas by amenity agglomeration, we can examine how different amenities tend to co-locate with each other. The network graph bellow shows amenities linked by the probabilities that they are in the same area. This probability is calculated by a Spearman Rank Coefficient (rho).

Click enlarge to view full screen.

Area Similarities

Using a hierarchical clustering method, we have identified 12 different clusters and have named them based on the top occurring amenities in each. Although the clustering was applied only using amenity counts the results show a clear spatial pattern.

Hover over the Map to get variables associated with each area.

Time to study Parking and diversity Free time Park and parking Drive through Pray and study For flower lovers Touristy Eat or drink Eat and coffee Eat and eat Go shopping

What is the relationship between diversity and other variables?

The graph below shows the correlation between amenity diversity in each area (measured by Shannon entropy index) to public transport accessibility, job / residential population ratio and medium income.

Hover over the points to see their location on the map.

About

What is it?

placeSpace is an interactive online platform that enables users to visualize and explore amenity location and distribution in urban environments. Currently the platform only supports London, but it can potentially be extended to other cities.

The platform originates from the need to interpret and assess the complexity of cities in terms of agglomeration economies. Using a dataset summarizing the precise location of thousands of amenities in London we visualize spatial distribution and relationship of amenities. The platform expands on the concept of Amenity Space introduced by Cesar Hidalgo by building a co-location network and exploring how amenity distribution and diversity define areas and correlate to other urban variables such as accessibility, residential population, and income.

The visualizations can help citizens, policy makers and researchers answer questions such as:

Data

Open Street Map icon

Open Street Map

Amenity data for the analysis were extracted from Open Street Map (OSM) through Overpass API. The selected elements are composed of amenities, shops, tourism and leisure inside Greater London. The data excludes amenity types such as urban furniture and natural and historical amenities. After collecting and cleaning of the OSM data, a new category was added to amenities as to divide them in specific types such as Food and Drinks, Services, Shopping, Education and Entertainment, Government and Other.

Open Street Map icon

London Data Store

Data regarding number of jobs, residential population, medium household income and PTAL (transport accesibility levels) at ward level was retrieved from London Data Store. To map this data to the polygons identified by our clustering algorithm, an area-based volume-preserving method called overlay interpolation was used.

Methodology

Our analysis consists of two steps: first, we identify amenity agglomeration clusters and second, we use these clusters as unit of analysis to reveal urban amenity micro-agglomeration structure. We studied micro-agglomeration structure from two angles: one is to uncover pair-wise amenity co-location pattern and explore if dominant amenities effectively categorize clusters; the other is to examine if amenity diversity within each cluster correlates with demographic indicators and public transportation accessibility.

The methods employed are as follows:

In order to identify neighborhoods based on amenity location agglomeration, we have experimented with two clustering methods — the classic Density Based Spatial Clustering Algorithm with Noise (DBSCAN) and Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN). After, we define geographical areas by drawing the alpha-shape polygons that tightly fit around clustering points, we chose HDBSCAN to proceed for further analysis because its result best corresponds to the known areas in London.

To examine amenity co-location pattern, we calculated probability for finding a pair of amenities in the same area through Spearman’s Rank Correlation. To explore how these colocation patterns may be similar or different across the city a hierarchical clustering based on number of occurrences of each amenity type was applied.

Finally, for each cluster, we calculated Shannon’s Entropy Index to quantify its amenity diversity. Introducing urban profiles other than amenity — namely, medium house income, ratio of jobs versus resident population, and accessibility in terms of public transportation — we converted these census data in unit of ward to our identified clusters through overlay areal interpolation. Lastly, we explore correlation between amenity diversity and these characters.

Team

Mateo Neira

Mateo Neira

Eleftherios Sergios

Eleftherios Sergios

June He

June He

Mercedes Landa

Mercedes Landa