Open 0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q opened 2 years ago
This is currently not possible. The reason is:
Somewhere needs to be a mapping of variable to oeo class. Currently this happens in a modified version of the metadata of the datasets. An example can be seen here. So for example the column manufacturer
(line 285) is connected to the oeo class OEO:00000045
.
Since the data in the rows is not included in the table metadata (since the metadata contains only information about the columns, not the content), in the current way this is implemented there can't be a way to annotate the rows, only the columns.
This is currently not possible.
So it will be in the future? Is anybody working on that?
Also, finally an answer after two years ;).
We (@giannou and I) looked at how table-metadata-mapping/out.csv is linked to the OEO in oep_metadata/turbine_metadata.jsonld. But the wide format (one column per variable) is … awkward for us. For one thing, we have many variables. The data for the Fifth Assessment Report had 148, that for the Special Report on 1.5°C had 598, and our model currently reports 2267. For another, people that already work with IAMC-style data are used to different formats, and they would need to change their behaviour in some way to adopt the Databus. We would try to minimise the need for the former to maximise the chances for the latter.
So, is it possible to annotate data by row, not by column? For the example data below, what we would ideally have is a file linking e.g. the variable
Primary Energy|Gas|w/ CCS
to the appropriate class in the OEO.