For an example of clustering with relational constraint I decided to construct a data set about European countries from CIA World Factbook and possibly extend it to world countries. I first tried to get shape files for Europe. It was not so easy - there are shape files for NUTS at EuroStat. Finally I found a Map of European Countries. A big problem with this map is that it does not contain Cyprus (, Vatican and Kosovo).
From shape files we can get using R the basic neighbors relation (see Neighbors Europe). This relation should be manually extended with missing countries.
As an alternative, I decided to find a good world map (World maps) and cut out the European countries. I checked two solutions proposed on the web: Map from WorldMap in R / 1 and Map from WorldMap in R / 2.
After some iterations my list of European countries converged to the following 57 countries:
> load('C:/Users/batagelj/Downloads/data/CIA/CIAeurope.Rdata')
> E2$alpha2
[1] "AL" "AD" "AT" "BE" "BA" "HR" "CZ" "DK" "EE" "FI" "FR" "DE" "GI" "GR" "GG" "HU" "IE" "IM" "IT" "JE"
[21] "LV" "LI" "LT" "LU" "MK" "MT" "MC" "ME" "NL" "NO" "PL" "PT" "SM" "RS" "SK" "SI" "ES" "SE" "CH" "GB"
[41] "AM" "AZ" "BY" "BG" "FO" "GE" "IS" "SJ" "MD" "RO" "SJ" "TR" "UA" "RU" "CY" "VA" "XK"
They are all included in the constructed neighbors network. Unfortunately I was unable to find a map containing all of them. Besides this also data on selected variables are not available for all of them. Therefore some of them will need to be removed in the construction of relational data used in clustering. Fortunately the problematic countries are “on the border”.
Extracting European data from CIA
Analysing European data from CIA
See also related notes: Colors in R; Maps in R; World map in R; Displaying incomplete data on the map; Coordinates; Unicode symbols on maps.