Covid_19: TG provides CSV via opendata.swiss

[metaodi]

TG data is available here: https://opendata.swiss/de/dataset/covid_19-fallzahlen-kanton-thurgau

[baryluk]

Interesting.

https://statistik.tg.ch/public/upload/assets/94501/COVID19_Fallzahlen_Kanton_TG.csv follows exactly our CSV format (it is using semicolons as separator).

It also has more data than us. It goes back to 2020-03-05.

However, looking at some recent values:

date;time;abbreviation_canton;ncumul_conf;ncumul_hosp;ncumul_ICU;ncumul_released;ncumul_deceased;source
...
05.04.2020;;TG;219;30;13;15;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552
06.04.2020;;TG;228;32;13;15;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552
07.04.2020;;TG;236;35;13;15;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552
08.04.2020;;TG;255;35;13;15;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552

But on website I see this: https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552

Aktuelle Fallzahlen im Kanton Thurgau

    Anzahl bestätigter Fälle: 255
    Hospitalisiert: 27
    davon auf der Intensivstation:  12
    Verstorben: 8

Stand 9.4.20

Note that hospitalized number on the website is 27, but according to csv file there were 30 since 2020-04-05 (or even 2020-04-02), and today there is 35.

Similarly ICU numbers seems off.

Would be good to understand why there is discrepancy.

https://statistik.tg.ch/public/upload/assets/94502/COVID19_Fallzahlen_Kanton_TG_Geschlecht_Alterskategorie.csv looks also good, and has extra breakdowns (age group and sex).

[baryluk]

Another plus of the CSV file, it provides information about released patients.

[metaodi]

I will check with them. Thanks for the review.

[dkoltg]

Hi, thanks for the question. There is a mistake in the version of this morning. I will update the revised dataset in the next few minutes. The column to hospitalised patients is the currently hospitalised number.

[baryluk]

@dkoltg AFAIK, we don't track explicitly cumulative hospitalizations in this repo. There is new_hosp that can be made to track cumulative hospitalizations, but it is incremental itself. But it should be possible to calculate new_hosp from ncumul_hosp in this CSV file from TG.

[metaodi]

I asked @dkoltg to provide the CSV with the new non-cumulative fields current_hosp and new_hosp.

[dkoltg]

I will be able to add the new_hosp numbers for the whole time after easter, sorry for the delay. The dataset is now update with columns to the current_hosp and current_icu...and newly hospitalized since the day before yesterday. Sorry for the inconvenience.

[metaodi]

@dkoltg nothing to be sorry about. Thanks for providing the CSV and Happy Easter! :rabbit:

[baryluk]

@dkoltg Thank you!

I notice, that you also left ncumul_hosp column, but it is empty. Would be great to actually have this data in the CSV. You had it there just hour ago.

Now it is not possible to calculate fully cumulative hospitalizations.

If it could look something like this:

06.04.2020;;TG;228;32;24;;;12;25;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552;;;;;;
07.04.2020;;TG;248;35;27;3;;12;25;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552;;;;;;
08.04.2020;;TG;255;35;27;0;;12;25;8;https://www.tg.ch/news/fachstab-gesundheit-coronavirus.html/10552;;;;;;

that would make it a bit better.

I also noticed that values for ICU changed. So it now looks it is not possible to get fully cumulative ICU numbers either?

The fact that we don't track these numbers in this repo (yet), doesn't mean we don't want them to be provided. It would be great to have them!

[dkoltg]

Hi @baryluk
Thanks to your question, I discovered some mistakes in the dataset. I updated it just now, without columns for cumulative numbers for hospitalisation and icu as I don't have the data at the moment. I will try to add all numbers for new hospitalisations and - if possible - also for new icu-cases for the whole period. However, that won't be before easter. Thanks for your patience.

[baryluk]

@dkoltg

Thanks for the clarification. No worries at all. We will revisit after Easter then.

[dkoltg]

@baryluk
Thanks for your help to improve the dataset! Happy easter :)

[dkoltg]

Hi, I had to modify the TG-dataset because there were some differences between the numbers the Gesundheitsamt published and the dataset on opendata.swiss. I will fill up the missing cells for current_hosp, current_icu and new_hosp after easter as soon as I will have received all the necessary data from the correspondent person.

[baryluk]

@dkoltg Thank you. new_hosp or cumul_hosp (total real cumulative since the start) are both find for us. I personally prefer the cumulative, but any way should be fine.

Do you know how often your dataset on opendata.swiss / github, is going to be updated? If it is going to be back and forth everyday, between you and others, I am not sure how sustainable it is. If you could ask them to send the data regularly by email, that could work I guess. What do you think?

Have a good day.

[dkoltg]

@baryluk Hi, I am not sure I understand your question. I update our dataset everyday before noon with the number that are reported by our heath office around 10 am. The first dataset I got was the one from 4/9 and I hope to get the older datasets soon, so that I can fill the empty cells for new_hosp. As I try to update the data in the same way every day (apart from the first struggles I had with them as the topic was completly new for me), I guess the results should be sustainable. Or what do you mean?
Best wishes,

[baryluk]

@baryluk Hi, I am not sure I understand your question. I update our dataset everyday before noon with the number that are reported by our heath office around 10 am. The first dataset I got was the one from 4/9 and I hope to get the older datasets soon, so that I can fill the empty cells for new_hosp. As I try to update the data in the same way every day (apart from the first struggles I had with them as the topic was completly new for me), I guess the results should be sustainable. Or what do you mean?
Best wishes,

Yes, that sounds great. I was just checking.

[metaodi]

@baryluk are you planning to work on this? Or is this up for grabs?

[metaodi]

This is currently not needed as TG will start to update this repo directly. For now the current scraper acts as a fallback.