filmov
tv
AI in German Official Statistics I Data for Policy 2024
Показать описание
"AI in German official statistics - from first steps to recent challenges."
---------------------------------------------------------------------------------
OVERVIEW:
📋 While (partial) automation aims at the efficiency of statistics production and possibly opens up new possibilities of data processing, the aspect of quality must not be neglected, especially in official statistics.
---------------------------------------------------------------------------------
Submission: #5483
AUTHOR:
Florian Dumpert
Federal Statistical Office of Germany
---------------------------------------------------------------------------------
Follow Data for Policy here:
---------------------------------------------------------------------------------
#ai #statistics
Keywords: official statistics, machine learning, quality, efficiency, automation
---------------------------------------------------------------------------------
ABSTRACT:
Official statistics face a variety of challenges worldwide. Driven by the increased possibilities of obtaining information and the progress in information technology, the demand for information from politics, the economy and society on the most diverse subject areas of official statistics is increasing. In order to meet this demand adequately, the production of statistics must be further developed. This is not only a matter of making new data sources usable, but also of making processes more efficient. This applies in particular to steps in the area of data processing (GSBPM phase 5), such as classify and code, review and validate, and edit and impute. In many NSOs (national statistical offices), solutions for (partial) automation of the processing steps are therefore being tested and used. Without the use of such statistical machine learning methods, it would not even be feasible to carry out some statistics due to their high frequency for instance. The talk will highlight classes of examples of how machine learning improves the production of German official statistics.
While (partial) automation aims at the efficiency of statistics production and possibly opens up new possibilities of data processing, the aspect of quality must not be neglected, especially in official statistics. Bad quality reduces trust very quickly. Existing frameworks at the level of the United Nations, at the supranational level (e.g. for the European Union) or at the national level frequently and rightly consider general requirements for the statistical institution, the processes and the statistical products. However, a concretisation for special situations, such as the use of machine learning, is necessary. There are first international and national works that address this concretisation. The talk will highlight the challenges and how to deal with them using conceptual and operational examples.
---------------------------------------------------------------------------------
OVERVIEW:
📋 While (partial) automation aims at the efficiency of statistics production and possibly opens up new possibilities of data processing, the aspect of quality must not be neglected, especially in official statistics.
---------------------------------------------------------------------------------
Submission: #5483
AUTHOR:
Florian Dumpert
Federal Statistical Office of Germany
---------------------------------------------------------------------------------
Follow Data for Policy here:
---------------------------------------------------------------------------------
#ai #statistics
Keywords: official statistics, machine learning, quality, efficiency, automation
---------------------------------------------------------------------------------
ABSTRACT:
Official statistics face a variety of challenges worldwide. Driven by the increased possibilities of obtaining information and the progress in information technology, the demand for information from politics, the economy and society on the most diverse subject areas of official statistics is increasing. In order to meet this demand adequately, the production of statistics must be further developed. This is not only a matter of making new data sources usable, but also of making processes more efficient. This applies in particular to steps in the area of data processing (GSBPM phase 5), such as classify and code, review and validate, and edit and impute. In many NSOs (national statistical offices), solutions for (partial) automation of the processing steps are therefore being tested and used. Without the use of such statistical machine learning methods, it would not even be feasible to carry out some statistics due to their high frequency for instance. The talk will highlight classes of examples of how machine learning improves the production of German official statistics.
While (partial) automation aims at the efficiency of statistics production and possibly opens up new possibilities of data processing, the aspect of quality must not be neglected, especially in official statistics. Bad quality reduces trust very quickly. Existing frameworks at the level of the United Nations, at the supranational level (e.g. for the European Union) or at the national level frequently and rightly consider general requirements for the statistical institution, the processes and the statistical products. However, a concretisation for special situations, such as the use of machine learning, is necessary. There are first international and national works that address this concretisation. The talk will highlight the challenges and how to deal with them using conceptual and operational examples.