Levelling-up: local authority cluster analysis

Exploring similarities between local authorities 

As part of their research into subnational variation in the UK the Office for National Statistics have released experimental findings from a cluster analysis of local authorities in England. Cluster analysis is a machine learning technique that uses an algorithm (in this case K-means) to classify observations into groups based on their degree of similarity.

The ONS have used the data from their subnational indices explorer to group local authorities together. The results contain one headline classification based on a selection of headline metrics and then additional classifications using the indicators in different domains (education, health etc).

These classifications are not intended to influence levelling-up policy decisions nor are a judgment about the relative performance of a local authority. They are experimental findings that allow researchers and users to understand similarities between different local authorities and identify control groups for research purposes. The results can be explored in the interactive graphic below and you can also read more about the research on the ONS website.
 

North East findings 

The results are helpful for determining on which domains the seven North East local authorities are most like one another. For instance, all seven local authorities were grouped together into the same cluster in the economic domain, being placed in the ‘below median on all economic metrics cluster’. This is indicative of the common economic challenges facing the region.

Health and skills are the other areas where the North East showed a significant degree of convergence. Six of the seven local authorities were placed in the ‘far below median health on all metrics cluster’. Northumberland was the only exception due to it being closer to the median, although it was still below the median on all metrics. 
In terms of skills five of the seven local authorities were in the ‘above median number of apprenticeships and further education cluster’, while Northumberland was in the ‘slightly below average on all skills metrics cluster’ and Newcastle in the ‘above median on level-3 and above qualifications cluster’.

In other domains the region was placed in a greater number of clusters. In terms of wellbeing there was a three-way split. Northumberland, Newcastle and Gateshead were grouped together in an ‘above average wellbeing cluster’. Durham and South Tyneside were then grouped in an ‘about average cluster’, while North Tyneside and Sunderland were in the ‘below average wellbeing cluster’. The grouping for education was even more diverse, with the seven local authorities being placed in four different clusters.

All these clustering’s should be taken as indicative and may be subject to revisions. Yet these groupings do help identify where the North East has common regional challenges and as well as providing a way of identifying suitable comparator areas.