Create a map with R — rgeoapi, ggmap and ggplot2

Improve your mapping with rgeoapi, a package designed to access the french geographic database.

rgeowhat ?

Developped by Etalab, with La Poste, l’INSEE and OpenStreetMap, the GeoAPI API is a JSON interface designed to make requests on the French geographic database.

rgeoapi was developped to facilitate your geographic projects by giving you access to these informations straight inside R. With rgeoapi, you can get any coordinate, size and population of a French city, to be used in your maps.

To install :

rinstall.packages("rgeoapi")

Get coordinates

So, let’s imagine you have a dataset with only the names of the cities you want to map. Like:

villes <- data.frame(nom = c("Rennes", "Lorient", "Brest", "Vannes"), variable1 = c("a", "b", "c", "b"), variable2 = c("Un", "Deux", "Un", "Deux"))

nom	variable1	variable2
Rennes	a	Un
Lorient	b	Deux
Brest	c	Un
Vannes	b	Deux

To put these cities on a map, you’ll need their coordinates. For that, you can use rgeoapi!

library(rgeoapi)

library(plyr)
geo <- ldply(villes$nom, ComByName)

name	codeInsee	codeDepartement	codeRegion	population
Rennes	35238	35	53	211373
Rennes-le-Château	11309	11	76	58
Rennes-les-Bains	11310	11	76	258
Rennes-sur-Loue	25488	25	27	88
Rennes-en-Grenouilles	53189	53	52	117
Lorient	56121	56	53	57961
Brest	29019	29	53	139386
Brestot	27110	27	28	518
Esboz-Brest	70216	70	27	485
Vannes	56260	56	53	53032
Vannes-le-Châtel	54548	54	44	579
Pouy-sur-Vannes	10301	10	44	145
Saulxures-lès-Vannes	54496	54	44	363
Vannes-sur-Cosson	45331	45	24	589

surface	lat	long	X_score
5035	48.11	-1.679	1
1497	42.91	2.277	0.7636
1975	42.92	2.341	0.7533
547	47.01	5.855	0.6743
801	48.49	-0.5083	0.6239
1533	47.75	-3.378	1
4948	48.4	-4.502	0.7183
884	49.34	0.6783	0.6958
977	47.8	6.441	0.4919
3313	47.65	-2.749	1
1747	48.57	5.785	0.7384
1579	48.3	3.597	0.6873
1826	48.52	5.804	0.678
3558	47.72	2.202	0.6653

 

names(villes)[1] <- "name"
villes <- merge(villes, geo, by = "name", all.x = TRUE)

Create a map with ggmap and ggplot2

The ggmap package has been designed to produce background maps to be used with ggplot2. The “fast” function to create a map is qmap(x,y) - the first argument referring to the query (city / department / region …) and the second to the zoom level of Google map.

library(ggmap)

map <- qmap('Bretagne', zoom = 8)

Once you have a map object, you can use it as the first argument of your ggplot call, and then combine it with your usual geom:

map + geom_point(data = villes, aes(x = long, y = lat, color= variable2, size = surface))

 

Easy, isn’t it!