4 minutes

Spatial data is any type of data that directly or indirectly references a specific location. With its roots in geography and ability to map any point of reference however large or small, it brings essential context to data analysis. In today’s digital world spatial data unlocks understanding that isn’t always detectable from other numerical or graphical data and opens up better business prioritisation and planning decisions.

In the past spatial location intelligence was just the conclusion of evidence drawn from the amalgamation of field geography data sprinkled with photography – manual, error prone and subject to potential results bias. With the application of today’s technology, digital techniques and availability of much wider data sets, it would be easy to assume that geo data scientists now have-it-made, but issues of inconsistent data aggregation, data quality and provenance remain the nemesis of the data practitioner’s day. Whether issues are recognised at the point of data acquisition, later in the analysis chain, or worse, not recognised at all, the application of external data doesn’t always enhance the route to results.

With margins in business tight, the pressure to deliver spatial data analysis that fuels better commercial decisions has never been higher. All would agree that ‘more’ data is not the answer to the most pressing locational questions in both the private and public sectors, so what is the data of today that can be applied safely and swiftly to help firms plan ahead and gain a competitive edge?

At the recent Spatial Data Science Conference in London a panel session examined what kind of present and future external data could make the difference.
The panel was moderated by Raquel Carvalho, Partner Development at NIQ, alongside Clifford McDowell, Founder and CEO at Doorda, Phil Cooper Global Geospatial Lead at Amazon Web Services and Justin Maynell Senior Account Manager at Experian.

The Evolution of UK Data

Circa 10 years ago is regarded as the point in time when attention shifted to contemporary UK datasets. A government initiative to create a site called data.gov. that made available non-personal UK government data as open data, kick started interest, but this was short-lived as it was impossible to keep the site updated. It was possible however to look at various individual, official and at-source data sets such as local authority data, Companies House data and other official government departmental data.

Pulling together data across key locational sets such as properties and businesses was, and remains, a difficult task, just as understanding and navigating the complex licensing needs specialist knowledge. However harmonising data and creating ready-made indexes has now become the forte of some specialist data providers. Data Marketplaces have also evolved exponentially over this time.

Not just in the UK, but globally, land mapping data has also particularly benefited from 50 years plus of satellite information. On-demand cloud computing platforms now have imagery and coordinates-based data on-tap to fuel the plethora of IOT and AI based applications that command pin-point routing, planning and predictive modelling. Tools that apply land use data to define business activities have also evolved immensely in the last 10-15 years.

Meeting the challenges of creating high-quality products using Open Data

Single data sets don’t yield results, data overlay and aggregation is where the magic happens for all data scientists, spatial or otherwise. The challenge remains correctly combining, quantifying and checking, and accurately understanding, so the quality of initial data input remains a priority for all.

Verification before aggregation is the initial and primary route to creating high quality products using open data but, when curating UK data products, the challenges are numerous. In UK property for example, issues with matching unique addresses, properties which change address, lack of official address history and data acquisition lag all make it difficult to create data sets that are accurate and relevant to their task.

Reliably linking every unique UK property to core data sets enables the triangulation that ultimately makes data products relevant- Find those data providers, with the trusted data, the data knowledge and the skills, and your location based models will sing new answers!