Online Program

Use of R in Business Surveys at the Italian National Institute of Statistics: Experiences and Perspectives
*Giulio Barcaroli, Istat 
Diego Zardetto, Istat 

Keywords: R software, business surveys, selective editing, calibration, estimation, sampling variance

Over the last six years, R has been steadily gaining ground in Istat, since a strategic decision to limit dependence on proprietary technologies (like SAS) was taken. A migration activity of our critical IT tools from SAS to R was carried out (we can cite MAUSS-R for optimal sample allocation, and ReGenesees for the calculation of estimates and sampling errors), and new R packages were developed (e.g. SeleMix for selective editing). In particular, ReGenesees has been recently experimented on a number of structural business surveys, namely “Information and Communication Technology in Enterprises”, “Community Innovation Survey”, “Access to Finance” and “Labour Cost”. Moreover, in the “Survey on Agricultural Production Prices”, various ad-hoc procedures (editing and imputation, calculation of indices), formerly developed in proprietary technologies, have been successfully migrated toward R, in a complex architecture where data reside in an Oracle database and the overall application is web based. A successful test activity has also been carried out for data retrieval and processing from the register of active enterprises (4.5 million units), thus showing that R limitations in handling huge datasets can be overcome.