Home

Dev meeting 28/05/2021

Prototype

Needs ironing out and tying together

• define the metadata that we need for running into the workflow - everything that is in the {indicator} package TODO: developers to review the indicators in the {indicator} package and sign them off, here is the issue. The package can deal now with two data sources: manually imported or created datasets, which must conform the indicator s3 class (see vignette or via the get_eurostat_indicator() function which (hopefully!) brings the data to a similar format like the indicator class, but does not coerce it to an indicator. [The coersion to an R class has no value to the SQlite API, it is a way to force through a thorough schema check, and add some automatic updates, like date of creation, to the metadata in case of manual entries to the database.]
• iotables here is a very-very detailed enhancement issue if anybody wants to try her/his hand on this package and create environmental impact analysis. Also, there is the issue of implementing good coding practices. I did a lot, and currently the package is CRAN-ready, but some minor improvements could be made - if anybody wants to do that, go ahead, if not, I'll close this issue.
• bookdown - we need to work on the automated organisation of the metadata files inside the long-form docs TODO: Daniel to share his original workflow
• add a description of the improvements that we made compared to the original Eurostat datasets
• new CRAN release of {iotables} TODO: Daniel to make changes TODO: other devs to review the package and play with it to get their head around the env indicators

Strategy

• automation pulling the data from source and versioning it
• Leo: Is there a risk of having to manually amend datasets on a regular basis? What overhead are we expecting?
• we have to accept that there will be certain datasets that will be hard to harmonize programmatically, but there will be some that can be quickly corrected with a high level of certainty (e.g. regional codes, country borders, etc.)