Methods library of embedded R functions at Statistics Norway
Journal article
Published version
Permanent lenke
https://hdl.handle.net/11250/2647376Utgivelsesdato
2017Metadata
Vis full innførselSamlinger
- Artikler / Journal articles [426]
- Publikasjoner fra Cristin [138]
Originalversjon
Romanian Statistical Review. 2017, (4), 129-139.Sammendrag
Statistics Norway is modernising the production processes. An important element in this work is a library of functions for statistical computations. In principle, the functions in such a methods library can be programmed in several languages. A modernised production environment demand that these functions can be reused for different statistics products, and that they are embedded within a common IT system.
The embedding should be done in such a way that the users of the methods do not need to know the underlying programming language. As a proof of concept, Statistics Norway soon has established a methods library offering a limited number of methods for macro-editing, imputation and confi dentiality. This is done within an area of municipal statistics with R as the only programming language. This paper presents the details and experiences from this work. The problem of fi tting real word applications to simple and strict standards is discussed and exemplifi ed by the development of solutions to
regression imputation and table suppression. Keywords: Offi cial statistics, R; Common Statistical Production Architecture, Generic Statistical Information Model, Validation and Transformation Language, Imputation, Statistical disclosure control
JEL Classifi cation: C18, C88