⚠️There's a newer version (4.0.0) of this package. Take me there.

ggmap

ggmap is an R package that makes it easy to retrieve raster map tiles from popular online mapping services like Google Maps and Stamen Maps and plot them using the ggplot2 framework:

library("ggmap")
#  Loading required package: ggplot2
#  ℹ Google's Terms of Service: <]8;;https://cloud.google.com/maps-platform/terms/https://cloud.google.com/maps-platform/terms/]8;;>
#  ℹ Please cite ggmap if you use it! Use `citation("ggmap")` for details.

us <- c(left = -125, bottom = 25.75, right = -67, top = 49)
get_stamenmap(us, zoom = 5, maptype = "toner-lite") %>% ggmap() 
#  ℹ Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.

Use qmplot() in the same way you’d use qplot(), but with a map automatically added in the background:

library("dplyr")
#  
#  Attaching package: 'dplyr'
#  The following objects are masked from 'package:stats':
#  
#      filter, lag
#  The following objects are masked from 'package:base':
#  
#      intersect, setdiff, setequal, union
library("forcats")

# define helper
`%notin%` <- function(lhs, rhs) !(lhs %in% rhs)

# reduce crime to violent crimes in downtown houston
violent_crimes <- crime %>% 
  filter(
    offense %notin% c("auto theft", "theft", "burglary"),
    between(lon, -95.39681, -95.34188),
    between(lat, 29.73631, 29.78400)
  ) %>% 
  mutate(
    offense = fct_drop(offense),
    offense = fct_relevel(offense, c("robbery", "aggravated assault", "rape", "murder"))
  )

# use qmplot to make a scatterplot on a map
qmplot(lon, lat, data = violent_crimes, maptype = "toner-lite", color = I("red"))
#  ℹ Using `zoom = 14`
#  ℹ Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.

All the ggplot2 geom’s are available. For example, you can make a contour plot with geom = "density2d":

qmplot(lon, lat, data = violent_crimes, maptype = "toner-lite", geom = "density2d", color = I("red"))

In fact, since ggmap’s built on top of ggplot2, all your usual ggplot2 stuff (geoms, polishing, etc.) will work, and there are some unique graphing perks ggmap brings to the table, too.

robberies <- violent_crimes %>% filter(offense == "robbery")

library("ggdensity")
library("geomtextpath")

qmplot(lon, lat, data = violent_crimes, geom = "blank", 
  zoom = 14, maptype = "toner-background"
) +
  geom_hdr(aes(fill = stat(probs)), alpha = .3) +
  geom_labeldensity2d(aes(lon, lat, level = stat(probs)), stat = "hdr_lines") +
  scale_fill_viridis_d(option = "A") +
  theme(legend.position = "none")
#  ℹ Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.
#  Warning: Ignoring unknown parameters: contour, contour_var
#  Warning: Ignoring unknown aesthetics: level

Faceting works, too:

qmplot(lon, lat, data = violent_crimes, maptype = "toner-background", color = offense) + 
  facet_wrap(~ offense)
#  ℹ Using `zoom = 14`
#  ℹ Map tiles by Stamen Design, under CC BY 3.0. Data by OpenStreetMap, under ODbL.

Google Maps

Google Maps can be used just as easily. However, since Google Maps use a center/zoom specification, their input is a bit different:

(map <- get_googlemap("waco texas", zoom = 12))
#  ℹ <]8;;https://maps.googleapis.com/maps/api/staticmap?center=waco%20texas&zoom=12&size=640x640&scale=2&maptype=terrain&key=xxxhttps://maps.googleapis.com/maps/api/staticmap?center=waco%20texas&zoom=12&size=640x640&scale=2&maptype=terrain&key=xxx]8;;>
#  ℹ <]8;;https://maps.googleapis.com/maps/api/geocode/json?address=waco+texas&key=xxxhttps://maps.googleapis.com/maps/api/geocode/json?address=waco+texas&key=xxx]8;;>
#  1280x1280 terrain map image from Google Maps; use `]8;;ide:help:ggmap::ggmapggmap::ggmap]8;;()` to plot it.
ggmap(map)

Moreover, you can get various different styles of Google Maps with ggmap (just like Stamen Maps):

get_googlemap("waco texas", zoom = 12, maptype = "satellite") %>% ggmap()
get_googlemap("waco texas", zoom = 12, maptype = "hybrid") %>% ggmap()
get_googlemap("waco texas", zoom = 12, maptype = "roadmap") %>% ggmap()

Google’s geocoding and reverse geocoding API’s are available through geocode() and revgeocode(), respectively:

geocode("1301 S University Parks Dr, Waco, TX 76798")
#  ℹ <]8;;https://maps.googleapis.com/maps/api/geocode/json?address=1301+S+University+Parks+Dr,+Waco,+TX+76798&key=xxxhttps://maps.googleapis.com/maps/api/geocode/json?address=1301+S+University+Parks+Dr,+Waco,+TX+76798&key=xxx]8;;>
#  # A tibble: 1 × 2
#      lon   lat
#    <dbl> <dbl>
#  1 -97.1  31.6
revgeocode(c(lon = -97.1161, lat = 31.55098))
#  ℹ <]8;;https://maps.googleapis.com/maps/api/geocode/json?latlng=31.55098,-97.1161&key=xxxhttps://maps.googleapis.com/maps/api/geocode/json?latlng=31.55098,-97.1161&key=xxx]8;;>
#  Warning: Multiple addresses found, the first will be returned:
#  !   55 Baylor Ave, Waco, TX 76706, USA
#  !   1301 S University Parks Dr, Waco, TX 76706, USA
#  !   HV2M+9H Waco, TX, USA
#  !   Bear Trail, Waco, TX 76706, USA
#  !   Robinson, TX 76706, USA
#  !   Waco, TX, USA
#  !   McLennan County, TX, USA
#  !   Texas, USA
#  !   United States
#  [1] "55 Baylor Ave, Waco, TX 76706, USA"

Note: geocode() uses Google’s Geocoding API to geocode addresses. Please take care not to disclose sensitive information. Rundle, Bader, and Moody (2022) have considered this issue and suggest various alternative options for such data.

There is also a mutate_geocode() that works similarly to dplyr’s mutate() function:

tibble(address = c("white house", "", "waco texas")) %>% 
  mutate_geocode(address)
#  ℹ <]8;;https://maps.googleapis.com/maps/api/geocode/json?address=white+house&key=xxxhttps://maps.googleapis.com/maps/api/geocode/json?address=white+house&key=xxx]8;;>
#  ℹ <]8;;https://maps.googleapis.com/maps/api/geocode/json?address=waco+texas&key=xxxhttps://maps.googleapis.com/maps/api/geocode/json?address=waco+texas&key=xxx]8;;>
#  # A tibble: 3 × 3
#    address         lon   lat
#    <chr>         <dbl> <dbl>
#  1 "white house" -77.0  38.9
#  2 ""             NA    NA  
#  3 "waco texas"  -97.1  31.5

Treks use Google’s routing API to give you routes (route() and trek() give slightly different results; the latter hugs roads):

trek_df <- trek("houson, texas", "waco, texas", structure = "route")
#  ℹ <]8;;https://maps.googleapis.com/maps/api/directions/json?origin=houson,+texas&destination=waco,+texas&key=xxx&mode=driving&alternatives=false&units=metrichttps://maps.googleapis.com/maps/api/directions/json?origin=houson,+texas&destination=waco,+texas&key=xxx&mode=driving&alternatives=false&units=metric]8;;>
qmap("college station, texas", zoom = 8) +
  geom_path(
    aes(x = lon, y = lat),  colour = "blue",
    size = 1.5, alpha = .5,
    data = trek_df, lineend = "round"
  )
#  ℹ <]8;;https://maps.googleapis.com/maps/api/staticmap?center=college%20station,%20texas&zoom=8&size=640x640&scale=2&maptype=terrain&language=en-EN&key=xxxhttps://maps.googleapis.com/maps/api/staticmap?center=college%20station,%20texas&zoom=8&size=640x640&scale=2&maptype=terrain&language=en-EN&key=xxx]8;;>
#  ℹ <]8;;https://maps.googleapis.com/maps/api/geocode/json?address=college+station,+texas&key=xxxhttps://maps.googleapis.com/maps/api/geocode/json?address=college+station,+texas&key=xxx]8;;>

(They also provide information on how long it takes to get from point A to point B.)

Map distances, in both length and anticipated time, can be computed with mapdist()). Moreover the function is vectorized:

mapdist(c("houston, texas", "dallas"), "waco, texas")
#  ℹ <]8;;https://maps.googleapis.com/maps/api/distancematrix/json?origins=dallas&destinations=waco,+texas&key=xxx&mode=drivinghttps://maps.googleapis.com/maps/api/distancematrix/json?origins=dallas&destinations=waco,+texas&key=xxx&mode=driving]8;;>
#  ℹ <]8;;https://maps.googleapis.com/maps/api/distancematrix/json?origins=houston,+texas&destinations=waco,+texas&key=xxx&mode=drivinghttps://maps.googleapis.com/maps/api/distancematrix/json?origins=houston,+texas&destinations=waco,+texas&key=xxx&mode=driving]8;;>
#  # A tibble: 2 × 9
#    from           to               m    km miles seconds minutes hours mode   
#    <chr>          <chr>        <int> <dbl> <dbl>   <int>   <dbl> <dbl> <chr>  
#  1 dallas         waco, texas 152474  152.  94.7    5428    90.5  1.51 driving
#  2 houston, texas waco, texas 298227  298. 185.    10301   172.   2.86 driving

Google Maps API key

A few years ago Google has changed its API requirements, and ggmap users are now required to register with Google. From a user’s perspective, there are essentially three ramifications of this:

  1. Users must register with Google. You can do this at https://mapsplatform.google.com. While it will require a valid credit card (sorry!), there seems to be a fair bit of free use before you incur charges, and even then the charges are modest for light use.

  2. Users must enable the APIs they intend to use. What may appear to ggmap users as one overarching “Google Maps” product, Google in fact has several services that it provides as geo-related solutions. For example, the Maps Static API provides map images, while the Geocoding API provides geocoding and reverse geocoding services. Apart from the relevant Terms of Service, generally ggmap users don’t need to think about the different services. For example, you just need to remember that get_googlemap() gets maps, geocode() geocodes (with Google, DSK is done), etc., and ggmap handles the queries for you. However, you do need to enable the APIs before you use them. You’ll only need to do that once, and then they’ll be ready for you to use. Enabling the APIs just means clicking a few radio buttons on the Google Maps Platform web interface listed above, so it’s easy.

  3. Inside R, after loading the new version of ggmap, you’ll need provide ggmap with your API key, a hash value (think string of jibberish) that authenticates you to Google’s servers. This can be done on a temporary basis with register_google(key = "[your key]") or permanently using register_google(key = "[your key]", write = TRUE) (note: this will overwrite your ~/.Renviron file by replacing/adding the relevant line). If you use the former, know that you’ll need to re-do it every time you reset R.

Your API key is private and unique to you, so be careful not to share it online, for example in a GitHub issue or saving it in a shared R script file. If you share it inadvertantly, just get on Google’s website and regenerate your key - this will retire the old one. Keeping your key private is made a bit easier by ggmap scrubbing the key out of queries by default, so when URLs are shown in your console, they’ll look something like key=xxx. (Read the details section of the register_google() documentation for a bit more info on this point.)

The new version of ggmap is now on CRAN soon, but you can install the latest version, including an important bug fix in mapdist(), here with:

if(!requireNamespace("devtools")) install.packages("devtools")
devtools::install_github("dkahle/ggmap")

Installation

  • From CRAN: install.packages("ggmap")

  • From Github:

if (!requireNamespace("remotes")) install.packages("remotes")
remotes::install_github("dkahle/ggmap")

Copy Link

Version

Down Chevron

Install

install.packages('ggmap')

Monthly Downloads

241,427

Version

3.0.1

License

GPL-2

Issues

Pull Requests

Stars

Forks

Maintainer

Last Published

November 3rd, 2022

Functions in ggmap (3.0.1)