FSO's Experimental statistics

Experimental statistics are produced using new methods and/or new data sources and are therefore in line with the FSO’s data innovation strategy and the Confederation's multi-annual programme for federal statistics. This site contains descriptions of the (pilot) projects currently being developed.

By publishing them we can involve users and partners at an early stage for both the development and consolidation of projects.

The aim of these statistical projects is to better meet users’ needs in terms of efficiency, quality and speed. However, these statistics still have the potential to evolve, especially regarding their methodology, which is still being assessed. For this reason they are clearly marked as experimental and carry a logo that can easily be recognised.

Published statistics

Go to Small area estimation (communes) of economic activity rate in the structural survey

Small area estimation (communes) of economic activity rate in the structural survey

The structural population survey provides important information on the population, including information about work. The whole purpose of Small Area Estimation is to push the boundaries imposed by standard methods.

The study showed that it is possible to obtain reliable estimates for both annual economic activity rates for communes that had a sample of at least 100 people.

Pilot projects within the data innovation strategy

On 21 November 2017, the FSO published its data innovation strategy .

This document is the FSO’s first response to the wider subject of digitalisation. More specifically, it focuses on the application of complementary analysis methods (e.g. predictive analysis using advanced statistical techniques, data science and machine learning) that enable the current production of official statistics to be increased or completed. Five pilot projects have been chosen to implement this strategy and are in progress. Each project is described below.

Go to Project ”Area Statistics Deep Learning” (ADELE)

Project ”Area Statistics Deep Learning” (ADELE)

The FSO’s land use statistics are an invaluable tool for long-term land observation. Thisproject involves learning and mastering the use of artificial intelligence (AI) technologiesto eventually automate (even partially) the visual interpretation of aerial images in order todetect and classify changes.

Go to Project ”Automation of NOGA coding” (NOGAuto)

Project ”Automation of NOGA coding” (NOGAuto)

Automation of the coding of the economic activity of enterprises using Machine Learningmethods applied to data already available within the FSO (data from surveys, descriptionsin the commercial register, keywords, explanatory notes for classifications etc.) to supportcoding.

Go to Project ”Machine Learning SoSi” (ML_SoSi)

Project ”Machine Learning SoSi” (ML_SoSi)

The clustering of typical prospective trajectories patterns concerning the receipt ofbenefits in the social security system as well as employment and the estimation of clusteraffiliation through the use of individual characteristics and retrospective process dataapplying a machine learning approach.

Go to Project ”Data validation with machine learning”

Project ”Data validation with machine learning”

The aim of this project is to extend and speed up data validation in the FSO by means of machine learning algorithms and at the same time to improve data quality.

Go to Project ”Evaluate the potential of small area estimation methods for the Job Statistics” (JOBSTAT)

Project ”Evaluate the potential of small area estimation methods for the Job Statistics” (JOBSTAT)

A project team of 15 people evaluates the potential of small area estimation methods forthe Job Statistics. The aim is to produce reliable estimates of the total number of jobs andFTEs for cantons, major towns and NOGA levels that were not anticipated in the sampleplan.