Integrated statistical indicators from Scottish linked open government data.

Area of Science:

• Data Science
• Public Sector Informatics
• Machine Learning Applications

Background:

• Open Government Data (OGD) offers significant potential for Machine Learning (ML) when integrated.
• Statistical indicators often exist with varying temporal, spatial, and measurement granularities, hindering direct ML application.
• Linked Data technologies and SPARQL queries can aid in data integration but require careful handling of data heterogeneity.

Purpose of the Study:

• To describe an integrated dataset of statistical indicators retrieved from the Scottish OGD portal.
• To facilitate the application of Machine Learning methods on heterogeneous OGD.
• To create a compatible dataset with matching temporal and spatial characteristics for enhanced analysis.

Main Methods:

• Utilized SPARQL queries and manual search to retrieve statistical data from the official Scottish data portal.
• Focused on identifying and integrating datasets with compatible temporal and spatial granularities.
• Compiled a dataset of 60 statistical indicators across seven categories.

Main Results:

• A consolidated dataset of 60 statistical indicators from seven categories (e.g., health, housing, crime) was created.
• The data pertains to 6,976 '2011 data zones' in Scotland, with a reference year of 2015.
• The resulting dataset is prepared for direct use by researchers, students, policymakers, and journalists.

Conclusions:

• The integrated dataset enhances the usability of Open Government Data for Machine Learning.
• This resource enables diverse social, business, and research scenarios through advanced data analytics.
• Facilitates novel insights and solutions by leveraging harmonized public sector data.