Meeting notes - March 12th, 2024

[Chris-ECE]

Machine learning

The following proposed text was proposed regarding machine learning, and amended to include mention of processing methods (as well as analytical methods), In the “applicability” section, after para 18:

In recent years, there has been increasing use of machine learning techniques in statistical production, where they have the potential to improve both the production process itself and the outputs of the process. Machine learning techniques can be considered in the context of processing and analytical methods more generally, alongside more traditional statistical methods, or methods specific to a particular type of data such as (geo-)spatial analysis, and references made to “analysis” within GSBPM may potentially include analysis using machine learning, depending on the context.

Action: The discussion of this text resulted in an agreement by InKyung to make a new suggestions and to then follow up on this by email.

GeoGSBPM

The following proposed text was proposed as a jumping-off point for work on GeoGSBPM (at the end of paragraph 16):

GSBPM is applicable to the processing of geospatial data, and contains a number of specific references to it. For those who are only interested in examining the role of GSBPM in relation to geospatial data, a useful resource is the work on GeoGSBPM, undertaken under the auspices of the HLG-MOS, which can be found here: https://statswiki.unece.org/display/GSBPM/GeoGSBPM

• InKyung suggested GeoGSBPM is not so much about geospatial data, but more about the infrastructure. Florian suggested also referencing the version of GSBPM for administrative data, however, this is an example from one country that is not internationally endorsed.
• Edgardo suggested re-wording para 16 to reflect design of processes, which was done. Reference to management was also made (following suggestion by Antonio), and a reference to processing geospatial data in a statistical context, including a footnote about the work on GeoGSBPM with a link to further info.

Relationship Management

• InKyung formulated some extra text for subprocess 2.3, to reflect the importance of understanding respondents and incorporating user perspectives in the context of declining response rates.
• Edgardo suggested ensuring this text remained relevant in future, and the text was reworded to mention the importance of understanding respondents and incorporate the user perspectives in the design of data acquisition, including the design of fieldwork operations and respondent engagement in case the survey collections.
• Some of the language in 2.3 still refers to collection of data, rather than acquisition. Nadia was happy with acquisition whereas Edgardo and Antonio preferred collection.
• Juan suggested referring to sources and perspectives in design of acquisition.
• Edgardo pointed out that the "collect" terminology is still present in the text descriptions, and asked whether the decision to replace that term with "acquisition". InKyung said it will be sent to countries for consultation, and used depending on their reaction to that change.

Action: To continue the discussion about collect/acquire terminology on github.