Re: #correspondencetables header format #correspondencetables
Massimo Pizzol
When I tried to create metadata yesterday for one of the tables (isic4 to Nace2 I think) that table was in the format you describe. The problem was that some columns had mixed data types (e.g. strings and numbers) and the infer function got it wrong so I had to manually fix this. So I would suggest to keep data types homogeneous within the same column – if possible!
Not sure how to deal with missing values but I don’t think using e.g. zero is a good option so I would also go for empty value. OK for separator and header.
From: <hackathon2019@bonsai.groups.io> on behalf of "miguel.astudillo via Groups.Io" <miguel.astudillo@...>
Dear all
For consistency I think we should specify a bit more the format of the final tables
Separator: I’d propose “,” null values: An empty value seems reasonable to me, but I do not have a strong preference. Header: just one row, the first one
That should help the parser later on. What do you think?
Miguel
From: hackathon2019@bonsai.groups.io <hackathon2019@bonsai.groups.io>
On Behalf Of michele.derosa@...
Sent: 25 March 2019 18:21 To: hackathon2019@bonsai.groups.io Subject: Re: [hackathon2019] #correspondencetables header format
Agree. Isn t this also the current format of the tables?
|
|