BACK TO SCHEDULE REGULAR

Session 39

Data Visualisation for Geostatistics

5 June 2026
10:30 – 11:45
ŠIBENIK V

Presentation title
Watts at Work: Analysing Non-Residential Electricity Consumption using Smart Meter Data
The National Statistics Office (NSO) Malta receives non-residential smart meter electricity consumption data at account level.

Read more Read less

This data has been linked to the statistical business register to produce consumption outputs by industry, in line with EU reporting requirements. While these requirements have consistently been met, the data has not been utilised to its full potential due to missing account identifiers which limit the number of direct links across the data sources.

This paper will discuss the methodology developed to clean contract unique identifiers within the smart meter electricity data. The data sources used and results will also be presented. The aim of developing this procedure was to increase data linkage across the administrative registers and thus limit the estimation required. Improved linkage allows the NSO to expand its output on non-residential electricity consumption beyond the breakdown by industry. Estimates can be produced by NACE classes and other variables available in the business register, such as size class. Increasing the capability for microdata linking also facilitates the use of location coordinate data, thus allowing for spatial analysis. A selection of results will be presented by NACE section and size class. The spatial distribution will also be presented through a selection of maps at 1km2 grid cells.

Prior to the implementation of the procedure, 80% of total non-residential electricity consumption was linked to the business register. This has increased to 92%. Although a complete link with the business register has not yet been achieved, the reduction in unlinked records improved the quality of the data. This has allowed for an expansion in the indicators that can be produced in terms of economic activity and geographical distribution

Main author / Presenter
Sarah Micallef
National Statistics Office

Read more Read less

Sarah Micallef is a young statistician from the island of Gozo in Malta. A keen interest in mathematics led her to enrol in a Bachelor of Science degree in Mathematics and Statistics and Operations Research at the University of Malta in 2019. After graduating in 2023, she joined the National Statistics Office Malta. She currently works in the Regional, Geospatial, Energy and Transport Statistics Unit, specifically focusing on energy statistics. In her role as statistician, she is involved in collection, synthesis and compilation of data from a variety of sources relevant to the domain.

Presentation title
From addresses to coordinates: Leveraging national address registers and open data for high-quality geocoding at scale
The rapidly increasing availability of geospatial data presents new opportunities and challenges for the development of official and experimental statistics.

Read more Read less

Climate risk assessment represents a key application area, where integrating climate model outputs with high-quality location data is essential for assessing population and economic exposure to natural hazards. Accurate addresses are key to locating populations, buildings, and economic assets exposed to risk. Yet their heterogeneous, often non-standardised formats across Europe pose substantial challenges for geocoding. In addition, freely available bulk geocoding services typically do not meet the scale and confidentiality requirements that are essential when working with sensitive financial and economic data.

This paper presents a methodological framework to derive geocoordinates using publicly available address datasets - national address registers and OpenStreetMap (OSM) - while maintaining computational efficiency. The procedure begins with extensive string cleaning and normalization, combining country specific rules with the open-source library libpostal. Matching proceeds via a waterfall strategy: records are first matched using the most granular information (house number, street name, postal code and city), after which progressively less detailed criteria are applied to unmatched records, down to matching with postal codes only. The algorithm was successfully applied to over 12 million EU-based firms while preserving confidentiality, hence addressing a critical challenge in integrating spatial precision with sensitive financial business data.

Two alternative sources of geocoordinates are evaluated: OpenStreetMap (OSM) and national address registers. Integrating administrative data markedly improves both the coverage and the precision of geocoding. With the proposed procedure, 55% of entities are matched at the highest precision level compared to 9% using OSM alone. The results highlight the critical role of national address registers, combined with systematic cleaning and standardisation, in achieving high geocoding quality. While such administrative data sources are rarely publicly available, enabling access to address registers maintained by National Statistical Institutes presents a significant opportunity to enhance European statistics.

Looking ahead, we propose three directions for improvement: refining address normalisation, integrating advanced techniques and prioritising national address registers as the primary geocoding source.

Main author / Presenter
Eva Pereira
European Central Bank

Read more Read less

Eva Pereira is a Research Analyst at the European Central Bank, working in geospatial data analysis since September 2024. She is part of the team responsible for compiling the ECB climate indicators on physical and transition risks. Eva holds a postgraduate degree in Data Science and Business Analytics from ISEG University in Portugal, where she graduated as the top student of her cohort and later served as a tutor for the program’s subsequent edition. Her research interests include geospatial modeling, data-driven policy analysis, and advanced analytics techniques. Before joining the ECB, Eva contributed to several projects at Banco de Portugal.

CO-AUTHOR:

Presentation title
High-Resolution Territorial Statistics Using Data integration, Multivariate and Spatial Methods
The increasing availability of administrative data provides unprecedented opportunities for producing high-quality official statistics that are timely, detailed, and policy-relevant.

Read more Read less

This paper illustrates how integrates administrative sources with traditional statistical surveys to produce fine-grained, spatially and multivariately informed estimates of Italy’s economic and socio-demographic structure. Statistical registers, administrative data —including business, tax, and social security data—offer comprehensive coverage of firms, institutions and individuals, ensuring reliability and minimizing non-response. When combined with surveys on labor force participation, technology adoption, innovation, and research and development, these sources provide both structural and behavioral information essential for a nuanced understanding of local economies.

The analysis employs multivariate techniques to examine the relationships among economic activities, innovation, human capital, and socio-demographic characteristics such as age and education. In parallel, spatial statistical methods are applied to Local Labor Systems, which represent functional economic areas reflecting the social and economic interactions of residents. This dual approach captures both the interdependencies among variables and the spatial distribution of economic dynamics, allowing the identification of clusters, spatial autocorrelation, and heterogeneity in innovation, knowledge accumulation, and competitive capacity. Patterns often masked in aggregate statistics emerge clearly when multivariate and spatial dimensions are combined.

Methodologically, the study relies on microdata integration (to harmonize multiple sources and address coverage and consistency issues), multivariate modeling and geospatial statistical techniques to identify patterns and interpret relevant relationships across time and space. This integrated approach improves the accuracy, granularity and interpretability of official statistics, supporting evidence-based policymaking, regional development planning, and the assessment of territorial disparities.

The Italian experience demonstrates the transformative potential of data integration when combined with multivariate and spatial analysis. This framework reduces respondent burden, improves timeliness, and provides actionable insights for policymakers, researchers, and stakeholders. By fully exploiting multiple sources alongside advanced analytical methods, national statistical institutes can produce reliable, fine-grained, and spatially coherent statistics that underpin effective governance and targeted regional interventions.

Main author / Presenter
Stefano De Santis
Istat

Read more Read less

S. De Santis, PhD, Senior Researcher. Works on the development of multi-source databases and statistical techniques for microdata integration, economic analysis, and policy evaluation (firms and labour market), as well as the assessment of European Structural Funds. Adjunct Professor of Databases at “Sapienza” University of Rome, SAS Instructor for ISTAT internal training, and member of the ISTAT R Experts Committee.

CO-AUTHORS: