Covid_19: Missing data in `ncumul_released` field (should be carried forward)

[Pseudomanifold]

Thanks for providing this service!

I saw that, contrary to the description of the data frame, the ncumul_released column is missing data for certain cantons (for example AG: the number increases to 4, but then no more data is being reported). I suggest forward-filling data or making a mention of this in the README.

Let me know if I can support this in any way!

[fabian]

Unfortunately AG stopped recording number of patients released on 19.03.2020 as instructed by BAG:

Aarau, 19. März 2020 15.00 Uhr
LAGEBULLETIN Nr. 15 Coronavirus (COVID-19)
Gemäss der Weisung des BAG werden die Anzahl der Geheilten nicht mehr erfasst.

https://www.ag.ch/media/kanton_aargau/themen_1/coronavirus_1/lagebulletins/200319_KFS_Coronavirus_Lagebulletin_15.pdf

[metaodi]

My understanding is, that if we put a number in a column, that number was communicated by a canton for that date. If there is no information (like for AG as @fabian mentioned), the field is empty.
We have a validation step, to ensure values only stay the same or increase for a the cumulative fields.

But I agree, this should be documented.

[Pseudomanifold]

Thanks for clarifying! This behaviour makes a lot of sense. Let me know if I can help in documenting it or related behaviours.

[baryluk]

Yes, to an extent filling the gaps is to be done by consumer of the files. If the row has some empty field, the consumer is free to use some previous rows to fill it up, with all the caveats of this approach. The data (csv files in this repo) itself shouldn't be really filled / guessed, because it is not what the source is providing, and we would be giving a bad picture of what actually is going on.

[metaodi]

@Pseudomanifold please review #337, does that clarify things? Any feedback is appreciated.

[Pseudomanifold]

Yes, thanks a lot!