GLA data science identifies homes of multiple occupancy

Greater London Authority pilot project provides early step in boroughs working together to develop analytical capabilities

A data science pilot run by the Greater London Authority (GLA) has enabled a handful of boroughs to identify possible homes of multiple occupancy, according to the official leading the project.

Andrew Collinge speaking at conferenceAndrew Collinge (pictured), assistant director at the GLA, told the Socitm spring conference that it plans to share the details around London and that there are now plans for a project on air quality data.

He said the organisation began the pilot nine months, spent a month working out the definition of the problem then five bringing together six local authorities with workable data. The data science exercise took two weeks, following which housing and environmental inspection teams have been given lists of homes that could be under multiple occupancy without being registered.

Collinge said the early signs are that it has provided the councils with an improved ability to identity relevant properties.

Various elements of the project will be made available for other authorities, including the data sharing agreements, privacy impact assessments, algorithms. “It will indicate how machine learning systems are becoming more robust in handling real world messy data,” he said.

Openly available

“Local authorities will be able to take and use them in their own environment, so everything becomes openly available.”

“It’s about starting the conversation, it’s about standards that in the fullness of time we will all need.”

He added: “We want to bring together the 33 boroughs –  admittedly it’s not going to happen quickly or easily – so we can drive a data sharing culture. It’s about harnessing the power of data analytics and some of the partnerships that are going to help us do this.”

The next project will involve using data on air quality, such as the levels of PM10 and PM2.5 particulates and NO2 in the atmosphere.

“We need to understand how new configurations of cheaper sensor technology can fill in those gaps to allow us to properly monitor and measure air quality and adopt a range of interventions,” Collinge said.

He also emphasised the importance of authorities sharing what they had learned on issues on a regional and national scale to reduce the duplication of effort. Examples of this are likely to include preparing for the EU General Data Protection Regulation, and explaining to the public the future role of algorithms in decision-making.

Other plans for London include the establishment of stores for open data, analytical data and a platform for data from internet of things devices.