3.1 Points | geocode() and geom_point()

ggmap::geocode() is a nifty function that returns the latitude and longitude of a location. Regrettably, we can’t use it this time because of googles new API key rules.

• Use https://www.google.com/maps, or some other service, to find the longitudes and latitudes of some locations. Your favourite gym or restaurant for example?

In my example code, I show how to use a tibble of locations and geocode() to get coordinates for several locations and bind them together. Since we can’t use geocode() I also show how to create a tibble of locations, longitudes and latitudes directly.

• Create a tibble with location, lon and lat.

#Create a tibble of Visby's most important locations
visby.locations <- tibble(location = c("Mullbärsgården, Visby",
  "Visby Hostel, Visby",
  "Glassmagasinet, Visby"))

#Get the geocode, the latitude and longitude, of the locations
visby.geo <- geocode(visby.locations$location)

#Create a data.frame of the data for easier plotting
visby.places <- cbind(visby.locations,visby.geo)

#Create the tibble manually
visby.places <- tibble(
  location = c("Mullbärsgården, Visby",
     "Visby Hostel, Visby",
     "Glassmagasinet, Visby"),
      lon = c(18.29203,18.28998,18.28720),
      lat = c(57.64182,57.63851,57.63729))

• When you have your locations dataset ready, use geom_point() to add markers to your map for your locations. Feel free to use geom_text() to also add labels for your locations. Notice that you can treat the ggmap() function like ggplot().

ggmap(visby.watercolor.map, extent="device") +
  geom_point(data = visby.places, aes(x = lon, y = lat), color = 'red', size = 3) +
  geom_text(data = visby.places, aes(label = location), hjust=0, vjust=-1)

3.2 Pubs

What do they mean by pubs? Do they mean publications? Nope! We are going to plot actual pubs, bars & inns in the UK using a public dataset from https://www.getthedata.com/open-pubs.

Download the dataset pubs.rda from course materials.

Note: I have processed this data slightly, if you want you can see how below, but it is not important for this exercise.

#Read the dataset
pubs <- read.csv("open_pubs.csv", stringsAsFactors = F)

#Set column names
colnames(pubs) <- c("fsa_id","name","address","postcode","easting","northing",
  "latitude","longitude","city")

#Convert long/lat to numeric
pubs$latitude <- as.numeric(pubs$latitude)
pubs$longitude <- as.numeric(pubs$longitude)

#Remove any rows with NA values
pubs <- pubs[complete.cases(pubs),]

Use what you have learned so far to complete as many tasks as you have time for. There is some help code and example plots of the town “Oldham” below, but really try to do it yourself.

1. Load pubs.rda into R.
2. Subset the dataset to a city.
3. Get a map of the city with ggmap.
4. Plot the position of the pubs onto the map using geom_point.
5. Create another plot where you plot the density lines of pubs using geom_density2d.

• Optional: Feel free to select an appropriate maptype that isn’t so busy with text, I suggest the darker toner ones.

6. Fill the density lines by overlaying with stat_density_2d as well. Use parametergeom = "polygon".

• Tip: You can use scale_fill_gradient2 and scale_alpha to adjust colors and alpha.

---

Example code:

Task 4:

ggmap(oldham.map, extent="device") +
  geom_point(data = oldham.pubs, aes(x = longitude, y = latitude),
    color = 'red', size = 2)

Task 5:

ggmap(oldham.toner.map, extent="device", darken = .7, legend = "bottomright") +
  geom_density2d(data = oldham.pubs, aes(x = longitude, y = latitude))

Task 6:

ggmap(oldham.toner.map, extent="device", darken = .7, legend = "bottomright") +
  geom_density2d(data = oldham.pubs, aes(x = longitude, y = latitude)) +
  stat_density_2d(data = oldham.pubs, aes(x = longitude, y = latitude,
    fill = ..level..,alpha = ..level..), geom = "polygon", color = NA)

Task 6, tweaked:

ggmap(oldham.toner.map, extent="device", darken = .7, legend = "bottomright") +
  # Commenting geom_density2d, it is prettier without the lines =)
  #geom_density2d(data = oldham.pubs, aes(x = longitude, y = latitude)) +
  stat_density_2d(data = oldham.pubs, aes(x = longitude, y = latitude,
    fill = ..level..,alpha = ..level..), geom = "polygon", color = NA) +
  scale_fill_gradient2("Pubs", low = "white", mid = "yellow",
    high = "red", midpoint = 125) +
  scale_alpha(range = c(0.05, 0.30), guide = FALSE)