spmap resulting in blank spaces

[Layton Field]

I am trying to produce a world map with countries grouped into 3 categories. I can get the following command to draw a perfect map, which includes all countries. However, when I add any attribute, the resulting map drops about a dozen countries. Why?

This code works fine:
spmap using global_coord , id(id)

I start loosing countries with either of these codes:
spmap type using global_coord, id(id) fcolor(Blues) ndfcolor(red) clmethod(unique)
spmap type using global_coord , id(id) fcolor(Blues2) ocolor(BuRd)

Every country has complete data available for the variables involved, so why is spmap dropping countries when I include an attribute?

Thanks!

[Patricio Mosquera]

Are you sure that you uploaded your database?

use global_db.dta, clear  

spmap type using global_coord , id(id) fcolor(Blues2) ocolor(BuRd)

[Layton Field]

Yes, I double checked that I loaded the correct file. This is driving me crazy.

[Friedrich Huebler]

What do the maps look like? Section 12 of the FAQ explains how you can attach images to a post. Please include the exact commands that you used to create the maps.

[Patricio Mosquera]

You can found a map here:

http://www.naturalearthdata.com/http..._countries.zip

I use the following code

wbopendata, language(en - English) indicator(it.cel.sets.p2) long clear latest
rename countryname country
sort country
save telephone, replace

shp2dta using world_adm0, database("wdb.dta") coordinates("wcoord.dta") genid(id) gencentroids(stub) replace

use wdb, clear
rename NAME country

sort country
save "wdb.dta", replace
merge country using  telephone.dta
spmap  it_cel_sets_p2 using wcoord , id(id) fcolor(Blues2) ocolor(BuRd)

I get this graph

graph_telephone.gph

Graph.pdf

[Robert Picard]

You are not matching some countries because of country name differences. You have 2-digit ISO country codes in both sources so the following example uses those codes to match countries. There are a few duplicates in both files; I just drop the duplicates.

Also, the shapefile contains geographic coordinates (lat/lon). I show how to use geo2xy (from SSC) to convert from geographic to cartesian coordinates. The map will have the same proportions as Google Maps.

* I saved the dataset created by:
* wbopendata, language(en - English) indicator(it.cel.sets.p2) long clear latest
* into "wopendata.dta"

* drop two observations without a 2-digit ISO country code
use "wopendata.dta", clear
list if mi(iso2code)
drop if mi(iso2code)
isid iso2code
save "wopendata_fixed.dta", replace

* implement the same fix with the shapefile database
shp2dta using "ne_110m_admin_0_countries/ne_110m_admin_0_countries.shp", ///
    database("wdb.dta") coordinates("wcoord.dta") genid(id) replace
use "wdb.dta", clear
list name iso_a2 if iso_a2 == "-99"
drop if iso_a2 == "-99"
isid iso_a2
save "wdb_fixed.dta", replace

* You should never make a map using unprojected coordinates. Use -geo2xy-
* to convert to x,y using the same projection as Google Maps. To install,
* type: ssc install geo2xy
use "wcoord.dta", clear
replace _Y = -90 if _Y < -90    // fix invalid coordinates
replace _X = 180 if _X > 180    // fix invalid coordinates

geo2xy _Y _X, replace
save "wcoord_xy.dta", replace

* now that we have a unique identifier in both dataset, merge the wbopendata
* with the shapefile database
use "wopendata_fixed.dta", clear
rename iso2code iso_a2
merge 1:1 iso_a2 using "wdb_fixed.dta"

* these are countries that have no match
list countryname iso_a2 if _merge == 1    // from "wopendata_fixed.dta"
list name iso_a2 if _merge == 2         // from "wdb_fixed.dta"

* make a map of countries that have matched
spmap  it_cel_sets_p2 if _merge == 3 using "wcoord_xy.dta" , id(id) fcolor(Blues2) ocolor(BuRd)
graph export "it_cel_sets_p2.png", replace

Here's the resulting map

[Friedrich Huebler]

Layton, has your problem been solved?

[Layton Field]

I am still working on the problem. I will try Robert's approach today and see if I can achieve the desired results.

So far, here is what I am doing:

use "C:\mapping_data.dta", clear
spmap using global_coord, id(id)

However, when I add any variable using the following code, countries start to disappear. I don't understand how one map is perfect but the addition of any new variable causes chaos:

spmap type using global_coord, id(id) fcolor(Blues) ndfcolor(red) clmethod(unique)

I can't find any duplicate measures but maybe I am missing something. Like I said, I will try Richard's approach and see what happens.

Thanks!

[Robert Picard]

I did not notice that #5 was from Patricio and not Layton. I'm stumped as to why polygons disappear like that.

[Friedrich Huebler]

Layton, even your original map is incomplete. Greenland, Iceland, the Western Sahara and many islands in the Pacific, Atlantic and Indian Oceans are missing. Can you try a different shapefile?

[Layton Field]

Yeah, I am working on a new shapefile now.

[Layton Field]

There must have been a problem with the shapefile. It looks like everything is better now! Thanks so much for all the suggestions!

[Julian Schaper]

I have a similar problem:

I get a map where all countries appear although I have missing data for most countries. For a year in which I have data for at least 68 countries I get a map in which most countries do not appear at all. I cannot make sense of that at all.

My code is

use master_0901.dta, clear


*harmonize cnames
replace cname="Cyprus" if cname=="Cyprus (1975-)"
replace cname="Swaziland" if cname=="Eswatini (former Swaziland)"
replace cname="Ivory Coast" if cname=="Cote d'Ivoire"

keep if year == 2017

save maps_data, replace

use db, clear

*merge
merge m:1 cname using maps_data

tab cname if _m==2

drop _m


*SDG10
*wdi_gini
spmap wdi_nersf using coord, id(id) fcolor(Blues)  ndocolor(black)   ndpattern(solid)  clmethod(custom) clbreaks(0 20 30 40 50 60)  legorder(hilo)  leg(pos(9)) title("Gini Index", size(medium))

It worked perfectly for 10 other variables just now. Is there any obvious mistake I am overseeing?

[Julian Schaper]

My apologies. I found the typing error