In ggfortify: Data Visualization Tools for Statistical Analysis Results

library(knitr)
opts_chunk$set(fig.width=3, fig.height=3, fig.path='figures/map-', warning=FALSE)

This document explains plotting geospatial data using ggplot2 and {ggfortify}.

Plotting with `{maps}` package

{ggplot2} can load map data provided by {maps} and {mapdata} package via map_data function. Loaded data is automatically converted to data.frame, thus you can plot maps via ggplot as below.

library(mapdata)
library(ggplot2)

jp <- ggplot2::map_data('world2', 'japan')
class(jp)
head(jp)
ggplot(jp, aes(x = long, y = lat, group = group)) +
  geom_polygon()

{ggfortify} additionally allows to autoplot map instances as it is. You can specify geom and other options to controll the outlooks.

library(ggfortify)
jp <-  map('world2', 'japan', plot = FALSE, fill = TRUE)
class(jp)
autoplot(jp)

p <- autoplot(jp, geom = 'polygon', fill = 'subregion') + 
  theme(legend.position="none")
p

Also, maps package provides some geospatial data. Following example retrieves Japanese city locations from world's city locations. Then plot on the previous map.

cities <- get('world.cities')
cities <- cities[cities$country.etc == 'Japan', ]
head(cities)

p + geom_point(data = cities, aes(x = long, y = lat),
               colour = 'blue', size = 0.1)

Because map plot created by {ggfortify} has a setting of aes(x = long, y = lat), you don't have to specify x and y aethetics in this case.

p + geom_point(data = cities, colour = 'blue', size = 0.1)

Plotting with `{sp}` package

Also, {ggfortify} can supports geospatial instances defined in {sp} package. Actually some functions are defined in {ggplot2}. Following table shows where each function is defined.

class | fortify | autoplot -------------------------------|-----------------|---------------- Line | {ggplot2} | {ggfortify} Lines | {ggplot2} | {ggfortify} Polygon | {ggplot2} | {ggfortify} Polygons | {ggplot2} | {ggfortify} SpatialLines | {ggfortify} | {ggfortify} SpatialLinesDataFrame | {ggplot2} | {ggfortify} SpatialPoints | {ggfortify} | {ggfortify} SpatialPointsDataFrame | {ggfortify} | {ggfortify} SpatialPolygons | {ggplot2} | {ggfortify} SpatialPolygonsDataFrame | {ggplot2} | {ggfortify}

Following example creates SpatialPointsDataFrame of some Japanese city's populatios, then plot it via autoplot. Note that you geom is specified by the original {sp} class. SpatialPoints is plot using geom_point for example.

library(sp)
df <- data.frame(long = c(139.691704, 135.519711),
                 lat = c(35.689521, 34.686316),
                 label = c('Tokyo', 'Osaka'),
                 population = c(1335, 886))
coordinates(df) <- ~ long + lat
class(df)
autoplot(df, p = p, colour = 'red', size = 10)

Also, you can use other columns as aethetics.

autoplot(df, p = p, colour = 'red', size = 'population') +
  scale_size_area()

Plotting with other packages

autoplot has p keyword to take existing ggplot instance. Below example plots SpatialPointsDataFrame on the ggmap.

library(ggmap)
bbox <- c(130.0, 30.0, 145.0, 45.0)
map <- get_openstreetmap(bbox = bbox, scale = 47500000)
p <- ggmap(map)
autoplot(df, p = p, colour = 'red', size = 'population') + 
  scale_size_area() +
  theme(legend.justification = c(1, 0), legend.position = c(1, 0))