GreenSquareAccord

GreenSquareAccord (GSA) was formed by the merging of GreenSquare and Accord Housing Association in 2021. Individually, GreenSquare owned and managed approximately 12,000 properties across Oxfordshire, Wiltshire and Gloucestershire while Accord Housing Association provided and managed approximately 13,000 affordable housing properties, care and support to 80,000 people across the West Midlands.

The combined GSA Housing Association now plans to add 1000 new homes each year to create a 25,000 property organisation. 

Accord and GreenSquare merged for their shared values and ambitions for expanding and enhancing locally focused services to customers. The merger created a stronger and more resilient organisation that’s more ‘future-proof’ and ready to deal with future challenges. They plan to use the extra resources resulting from the merger to expand and improve local delivery of services, across our larger operating area.  

GSA’s aims are to improve services to customers; invest more in local communities; extend our care and support services; and build more affordable homes.  

How we solved the problem

We deployed the Sentinel Data Integration Platform onto the GSA environment in order to bring all legacy data sets together.

Using the initial output from the Data Platform, we then conducted a requirement gathering phase, whereby the combined GSA / Sentinel team:

• Identified business and data requirements for the new integrated dataset
• Established the quality control needs of the latest HACT data model
• Confirmed the validation principles to be deployed across all of the legacy datasets so that new and consistent Corporate Data Standards would be applied
• Investigated all legacy datasets and establish the most appropriate matching criteria to suite the various data quality scenarios uncovered

This investigation included matching address records between the various legacy data sets, but also against the PAF (Post Office Address File) that Sentinel provides as part of this Data Integration Platform.

Machine Learning

Due to the number of records and properties and matching scenarios in this project, we utilized Microsoft Azure Machine Learning Services to compliment the in-built functionality within Sentinel’s Data Platform. We used these Machine Learning services to perform 1.5 trillion comparisons to discover and analyze potential matches.

This AI approach, including Natural Language Processing (NLP) and ANN or Approximate Nearest Neighbour (ANN), automated and accelerated the efficiency of matching properties. By harnessing the power of the Azure Machine Learning Services, the matching across all legacy stock records to the standard Royal Mail PAF database was reduced from weeks of matching time to less than 3 hours. 

The Azure Machine Learning Service also reverse engineered matching criteria so that as matches are successfully identified and validated the past matches are updated according to the latest insights – effectively double-checking previous match and cleanse work with the latest knowledge. By using Azure Machine Learning Services as a powerful analytics and validation tool our matching process delivered better quality data, sooner.  

The benefits of using our data platform

From Phase 1 of the project, GSA achieved their target requirement of a consolidated Property Stock List from all available legacy data. This provided a single, most trusted list, containing the most complete property details, without duplicates or erroneous entries.

This consolidated Property Stock List is being used to manage and improved their data quality according to the new Corporate Data Standards, and also as the basis for operational and financial reporting.

The project team were able to confirm where legacy data met the latest HACT requirements and where gaps existed.

The project continues to build on this, to further validate legacy records and combinations of records to ensure compliance to current and future reporting and Golden Thread requirements.