Missing IDs for several features

[curran]

Greetings,

I noticed there are several instances where the id is -99, and wanted to share this visualization of missing IDs.

Perhaps it's a problem with the upstream data in Natural Earth, but nonetheless it's surprising that the UAE has this value as its ID.

The Natural Earth Changelog indicates the latest release was 2018-05-21.

Maybe a fresh run of npm version patch and npm publish would fix the issue? I see the world-atlas package has not seen a fresh release for some time now.

Many thanks.

[curran]

Related #25

[curran]

It appears these have been solved in https://github.com/fil/visionscarto-world-atlas by @Fil

[mbostock]

In general, I don’t have the bandwidth to maintain manual patches to the Natural Earth Data. But it does seem pretty egregious that Norway’s country code is missing. Here’s my own visualization as of Natural Earth Data version 4.1.0 (world-atlas 2.0.0)

Let me take a look at some patches…

[mbostock]

The features with missing identifiers are:

• Somaliland - no defined ISO-3166-1 code (numeric or alpha)
• Norway - 578
• Kosovo - no defined ISO-3166-1 code (numeric or alpha)
• Northern Cyprus - no defined ISO-3166-1 code (numeric or alpha)
• Indian Ocean Ter. - unsure (separate from British Indian Ocean Territory)
• Siachen Glacier - unsure

The name property was added in world-atlas 2.0.0, so that can be used as a fallback to identify countries with no ISO-3166-1 numeric code.

I will patch Norway in 2.0.1.

[mbostock]

Fixed Norway in 2.0.1. I don’t know what to do for the other cases. Related #19 we could include the ISO alpha codes, but that would not help in this case.

[Fil]

In https://github.com/nvkelso/natural-earth-vector/blob/master/50m_cultural/ne_50m_admin_0_countries.dbf some codes are used:

• KAS (for Kashmir / Siachen)
• KOS (Kosovo)
• CYN (N. Cyprus)
• SOL (Somaliland)
• IOA (Indian Ocean Territory)

I've extracted the lines for convenience:
https://gist.github.com/Fil/401928df84ef0a16346ddfc37fd739dd

[mbostock]

@Fil Yes, but those aren’t the ISO-3166-1 numeric codes, which we’re using as identifiers here. And they aren’t the ISO alpha codes either; I believe they’re essentially made-up by Natural Earth Data with no standardized meaning.

[Fil]

Yes, it’s problematic. I use them though because what other choice do we have? It’s better than nothing in practical applications, esp. when you have data for these territories. Le dim. 1 sept. 2019 à 17:01, Mike Bostock <notifications@github.com> a écrit :

@Fil <https://github.com/Fil> Yes, but those aren’t the ISO-3166-1 numeric codes, which we’re using as identifiers here. And they aren’t the ISO alpha codes either; I believe they’re essentially made-up by Natural Earth Data with no standardized meaning. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#23?email_source=notifications&email_token=AAABWWPASJUMK2XS74PFELTQHPKNTA5CNFSM4FXKJCC2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5UEDTQ#issuecomment-526926286>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAABWWPASLF4DE7Y25WFTP3QHPKNTANCNFSM4FXKJCCQ> .

-- -- Fil

[mbostock]

But unless your data happens to use Natural Earth Data’s identifiers (and I haven’t seen any examples of this in public datasets), you’re going to have to map the data manually anyway. In this recent world choropleth for example, the WHO keys the data by name. Many of these names match the Natural Earth Data, but about 30 of them I had to map manually due to discrepancies in naming.

https://observablehq.com/@d3/world-choropleth

[Fil]

you’re going to have to map the data manually anyway

True. However I prefer to map from French names to a simple 3-letter indicator, rather than from French names to English names. In my process (or at least in my mind), it helps clarify what is what: messy input values vs. identifiers.

[mbostock]

That‘s fair, but I still don’t think it makes sense for Natural Earth Data’s SOV_A3 to be the primary identifier, since it is just a convention used by Natural Earth Data and not a standard (as I understand it from this comment: “The following 4 columns are unique to the Natural Earth administrative-0 coding system, but somewhat follow the ISO 3 digit alpha codes”). #19 discusses adding additional identifiers as properties.