XClose

UCL Social Data Institute

Home
Menu

Mikaella Mavrogeni & James Cheshire present at GLA High Streets Data Service (HSDS) ‘Data Day’ 

30 January 2025

James and Mikaella photo

Mikaella Mavrogeni (PhD student in the Department of Geography) and James Cheshire (Director of the UCL Social Data Institute) presented research on ‘Data After Dark: Nightworker Classification’ at the High Streets Data Service (HSDS) ‘Data Day’ showcase on 10 December at City Hall. The event was a celebration of the data-enabled innovation and collaboration shaping the future of London’s high streets and was organised by the Greater London Authority

Mikaella and James spoke about their project that combines a variety of datasets to create a more comprehensive picture of the geography of night working in London, by creating the The London Night-Worker Area Classification. One of the datasets used is the BT footfall dataset for 2024 that gives us the total footfall of residents, workers and visitors by 350-metre hexagonal grid by a 3-hour window. The data was accessed through the High Street Data Service of the Greater London Authority, and the event showcased the different use cases of their products. 

In their talk Mikaella and James outlined the background of the classification they created. They used a combination of variables coming from the BT data, and location of businesses that are likely to be contributing to the night-time economy. The choice of business variables was informed by a qualitative survey carried out by Didobi. BT data helped provide a direct understanding of  where night-workers are and a strong argument that was made is that most available datasets, indirectly capture night-workers. The BT footfall dataset on the contrary, can directly capture night-workers at both a temporally and spatially granular scale, and therefore it is very important to be able to access it.

In the talk Mikaella and James show some visualisations of the BT dataset such as the one below. This shows the average number of workers recorded within each 350-metre hexagonal grid by 3 hour windows:

 BT dataset 1

The graphic below shows the geodemographic classification of interactivity where each cluster is represented by a distinct colour and is built at the LSOA (Lower layer Super Output Areas) level:

BT data

The classifications were produced by following the well-established methodology of kmeans also followed by the LOAC, OAC., LWZC. 7 clusters have been used above - Central London is picked out as one of the clusters, then the other clusters are formed in rings around central London. The variation of the variables that make up each of the clusters can be assessed to get an understanding of what each cluster is made up for.

To summarise clusters based on the concentration of night-workers in relation to the London average, we can look at the figure below.This maps each cluster separately and ranks them based on the concentration of night-workers in relation to the London average:

BT dataset

The overall objective is to put the final classification on the London Datastore. Mikaella and James are also in the process of writing an academic publication to check the robustness of the methodology. 

Links:
Data After Dark research
Greater London Authority
UCL Department of Geography
UCL Social Data Institute