Random Forest: Seven ways to visualize points on the map using ggplot2 (on the example of McDonald's in Europe)

Roman Kyrychenko

There are many approaches for visualization of point data. Some of them are unusual. In this article, I’ll try to show different methods for visualization of the same data on maps. For this, I’ll use only ggplot2 package, but for data preprocessing I will need other packages (for example, maptools, sf, getcartr etc.).

First of all, let’s take a look at how we can show the same geographical information. We have data, which consist of coordinates. This is data about locations of McDonald’s restaurants in Europe, which I collected from Open Street Maps (OSM). They include only long and lat column. Also, we can:

map these coordinates as points;
calculate the density of points;
set a nearby area to a point;
estimate how many points are on each polygon (=region, country, etc.);
compute the shortest path through these points.

I think this is not a complete list of what we can do with this data. Also, let’s get this data first.

Data collection

I described how we could get data from OSM in one of my previous posts. But in this post, I collected lines data (about cycle paths). Here I want to get data about McDonald’s locations in Europe. Therefore I need to receive points, not lines.

This can be done directly using osmdata package. I recommend running this script as Job (This feature is in the preview versions Rstudio).


require(osmdata)

# for limit
b <- combn(seq(-17, 50, 2), 2) %>% t()
b <- b[!duplicated(b[, 1]), ]

mcdonalds_points <- lapply(1:33, function(i) {
  cw <- opq(
    bbox = c(b[i, 1], 35, b[i, 2], 70),
    timeout = 300, memsize = 1073741824
  ) %>%
    add_osm_feature("amenity", "fast_food") %>%
    add_osm_feature(
      key = "name",
      value = c("McDonald's", "McDonalds"),
      value_exact = FALSE, match_case = F
    )
  cwmap <- osmdata_sp(cw)
  cat(i / 33, "\n")
  cwmap$osm_points
})

coords <- purrr::safely(function(x) {
  f <- dplyr::as_tibble(x@coords)
  colnames(f) <- c("lon", "lat")
  f
})

mcdonalds_points <- purrr::map_dfr(
  mcdonalds_points, function(x) {
    coords(x)$result
  }
) %>% distinct()

# save data into rds format
readr::write_rds(mcdonalds_points, "mcdonalds.rds")

Hurray! We collected all European McDonald’s locations, which is marked in OSM.

Now it is possible to visualize data.

1. Scatter map

The simplest way to visualize points on the map is scatter map. In ggplot2 we can do it using borders and geom_point.


require(ggplot2)
require(hrbrthemes)

scatterplot <- ggplot() +
  borders("world", xlim = c(-10, 50), 
          ylim = c(33, 71), size = 0.05, 
          fill = "black") +
  geom_point(data = mcdonalds_points, aes(lon, lat), 
             color = "white", size = 0.1) +
  labs(
    title = "McDonald's locations in Europe",
    subtitle = "Scatter map",
    caption = "Based on OSM data"
  ) +
  theme_ipsum(base_family = "PT Sans") +
  theme(
    panel.grid = element_blank(),
    axis.text = element_blank(),
    axis.title = element_blank()
  ) +
  coord_map(
    projection = "gilbert", 
    xlim = c(-10, 50), 
    ylim = c(33, 71)
  )

scatterplot

Seven ways to visualize points on the map using ggplot2 (on the example of McDonald’s in Europe)

Table of Contents

Data collection

1. Scatter map