first/last observed values inverted in 5 distributions

trias-project / ad-hoc-checklist

🍀 Ad hoc checklist of alien species in Belgium

https://trias-project.github.io/ad-hoc-checklist

MIT License

1 stars 2 forks source link

first/last observed values inverted in 5 distributions #41

Closed damianooldoni closed 3 months ago

damianooldoni commented 2 years ago

See the temporal distribution value for Flanders in https://api.gbif.org/v1/species/164300894/distributions: "2019/2018". It should be "2018/2019", right?

The same happens for other 4 distributions:

164300756 (2019-2018), Flemish Region: https://api.gbif.org/v1/species/164300756/distributions
168130912 (2020-2018), Walloon Region: https://api.gbif.org/v1/species/168130912/distributions
168131164 (2019-2018), Flemish Region: https://api.gbif.org/v1/species/168131164/distributions
164301110 (2019-2018), Walloon Region: https://api.gbif.org/v1/species/164301110/distributions

timadriaens commented 2 years ago

indeed, earlier year first

qgroom commented 2 years ago

What do I need to do to resolve this? Where do they come from?

damianooldoni commented 2 years ago

In the README.md, I read that the raw data are maintained in this Google Spreadsheet.

qgroom commented 2 years ago

I think this has been fixed. I don't see the problem records. Though I don't know it it has been republished to GBIF yet

damianooldoni commented 2 years ago

@LienReyserhove: do you have plans to republish this checklist? In any case, I have solved the issue with a patch in the indicators workflow: https://trias-project.github.io/indicators/01_get_data_input_checklist_indicators.html#41_Split_column_temporal

LienReyserhove commented 4 months ago

This issue is still valid, the inverted mapping is a consequence of the way we consider the date of last observation when this information is lacking in the source dataset (i.e. as the last publication date of the checklist, which is 2018 in this case ...). This checklist needs an urgent update anyway, so I'll take care of it and republish the dataset asap.

timadriaens commented 4 months ago

I believe we need to reconsider that pragmatic solution as actually I think it is better not to "invent" a date of last observation

LienReyserhove commented 4 months ago

We really need to make a decision about this, preferably as soon as possible as @sangovae en I are now updating the publication pipelines. This discussion keeps coming back in several other issues (e.g. #34 or #42) . In #42 , I suggested a consensus approach. @qgroom @damianooldoni @peterdesmet

LienReyserhove commented 3 months ago

This question is answered here: https://github.com/trias-project/ad-hoc-checklist/issues/42#issuecomment-2166039019