Russia is missing

[pschloss]

I noticed that the gapminder site appears to have data for Russia, but it is missing from {gapminder}. If there's another release of the package it might be nice to have these data included.

Pat

[jennybc]

If I search this repo for "russia", it comes up quite frequently in the data preparation code:

https://github.com/search?q=repo:jennybc/gapminder%20russia&type=code

I can't immediately see any sort of explicit reason why Russia fell out of the dataset. It would take some sleuthing: re-running parts of the data prep code while keeping a specific eye on Russia to see when/why it disappears.

And then there's the question of whether to add it. Maybe? But it's not entirely clear. This dataset has various flaws and out-of-date qualities. It has become a bit like mtcars or iris. Fixing issues has some value, but so does stability.

In any case, I think it's worth some study.

[TimBFrankland]

The former Soviet Union (FSU) countries (Latvia, Russia, Ukraine, Belarus, Lithuania, Georgia, Armenia, Kazakhstan, and Uzbekistan) included in the gapminder source data do not meet the criteria for inclusion into the gapminder data package.

Specifically, these FSU countries fail to reach the minimum threshold (n=12) of observations from 1950 to 2007. The gapminder data package includes year data in 5-year intervals.

If data users want to bring data from FSU countries into their analyses, the data is available in the 07_gap-merged-with-continent.tsv file within this repo. The continent for these former FSU countries is listed as ‘FSU’ in the .tsv file noted above and users may need to re-specify the continent to fit their analyses.

If data users want to incorporate these FSU countries into existing gapminder plot data, then they will need to add colors for the FSU countries not already in use in the 40_country-colors.tsv.

[jennybc]

Thanks @TimBFrankland for doing this post mortem analysis! This rings a vague bell, i.e. I once knew this would be a consequence of a data cleaning move I made. And I also didn't see a way around it, assuming I wanted to create a data set with no missing data.

I think changing anything would be a net negative at this point, given how gapminder is used. But I'm pleased to have this documented.