Service: Trade statistics - Classification system that automatically new data points as relevant

Responsible organisation: Statistisches Bundesamt (Central-Government)

The starting point is a classification problem: In order to decide whether a data point (e.g. a craft enterprise) is to be included in a statistic, it must be classified as relevant. (e.g. a craft enterprise) is included in a statistic, it must be classified as relevant or not. or not relevant. Up to now, this has been done manually. However, this is time-consuming and is also prone to errors. Furthermore, there are only scarce personnel capacities. On the basis of previously determined classification patterns for units (information from business registers), a classification system is trained that automatically classifies new automatically classifies new data points as relevant or not. Case handlers are given the same time window as before to process the now pre-sorted cases. The classified data points (craft enterprises relevant/not relevant for the statistics). statistics)

Additional information

Source Open Innovation Regione Lombardia
Web site https://www.researchgate.net/publication/329945338_Machine_Learning_in_Official_Statistics
Start/end date 2018 -
Still active?

Related cases