Fully relationally normalized DB about coviddata HOT 3 OPEN

This is a great idea. The current format of the CSVs is based on the JHU CSVs' format to reduce switching costs, but a format in which the dates are row values instead of column headers would be preferable for many use cases.

Since there are three models (Countries, Regions, and Places), we would probably have either three CSVs (including location columns) or six CSVs (using normalization, as you describe). Or we could provide both formats. These would be provided in addition to the existing format.

Any thoughts on these options, relative to the two-CSV option that you've described?

from coviddata.

So I posted this issue last night I got too frustrated with not finding anything out there and went ahead and made a repo for the stuff I had already put together.

https://github.com/covid-db/covid-db

I definitely think we should be pooling efforts and avoiding repeated work. Not sure how best to merge what you're doing with what I have there. Your goal of more closely matching JUH's format makes sense too. Open to ideas about how to efficiently collaborate. I'm really interested in expanding to apply the same relational normalization approach to lower level data gathered from individual state, county, or even individual hospital data. Maybe take a look and see what you think?

To answer your question, I had been thinking that it might be good to provide the relational tables at different levels of granularity. Perhaps one for country-level, one for region (state) level, and one for county (or whatever the more internationally generalized term is). It might be ok to combine the country and region level into one table because it seems like region-level data is pretty widely available. Smaller granularities than that don't seem to be widely available, so they might need to be in separate files.

from coviddata.

Hi,

This project converts the Johns Hopkins CSV files to a fully normalised database:

https://github.com/MelbourneDeveloper/COVID-19-DB

Christian

@mightybyte

from coviddata.