In this blogpost I provide a coding example in R for how to create a map-based scatterplot using the deckgl package. This can come in handy when visualising data with some spatial aspect. For example you might want to visualise the geo-spatial distribution of certain property clusters.

Before I can apply the deckgl package’s functionality I need a geocoded dataset, i.e. a dataset which contains information about longitude and latitude coordinates of the property of interest. For this I will use a geocoding function applying the Open Street Map API. I found the function on datascienceplus.com.

# osm geocoder 
# source: https://datascienceplus.com/osm-nominatim-with-r-getting-locations-geo-coordinates-by-its-address/
osm_geocoder <- function(address = NULL)
{
  if(suppressWarnings(is.null(address)))
    return(data.frame())
  
  tryCatch(
    d <- jsonlite::fromJSON( 
      gsub('\\@addr\\@', gsub('\\s+', '\\%20', address), 
           'http://nominatim.openstreetmap.org/search/@addr@?format=json&addressdetails=0&limit=1')
    ), error = function(c) return(data.frame())
  )
  
  if(length(d) == 0) 
    return(data.frame())
  
  return(data.frame(lon = as.numeric(d$lon), lat = as.numeric(d$lat)))
}

Next, I need to initialize the data I want to plot. I stored a list of cities in a separate csv-file. I thus read in this file and convert it into a dataframe. I then use the geocoding function to geocode all the cities in my data frame. In addition I add normally distributed values to the “entries” and “exits” column; required for determining e.g. circle design in the scatterplot.

# ensuring that required packages are loaded
library(deckgl)

## deckgl 0.1.8 wrapping deckgljs 6.2.4
##   Documentation: https://crazycapivara.github.io/deckgl/
##   Issues, notes and bleeding edge: https://github.com/crazycapivara/deckgl

library(magrittr)
library(jsonlite)
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

# setting up the data frames 
scatter_data_df_1 <- data.frame(matrix(nrow=30,ncol=6))

column_names <- c("name","code","address","entries","exits","coordinates")
colnames(scatter_data_df_1) <- column_names

city_list_1_df <- read.csv("city list 1.csv",header = FALSE, stringsAsFactors = FALSE)

# geocode cities into longitude and latitude
for(i in 1:nrow(city_list_1_df)){
  dum_coord <- osm_geocoder(toString(city_list_1_df$V1[i]))
  scatter_data_df_1$name[i] <- paste0("city liste 1 : ",i)
  scatter_data_df_1$code[i] <- c("CL1")
  scatter_data_df_1$address[i] <- toString(city_list_1_df$V1[i])
  scatter_data_df_1$entries[i] <- as.integer(rnorm(1,mean=3000,sd=1000))
  scatter_data_df_1$exits[i] <- as.integer(rnorm(1,mean=3000,1000))
  scatter_data_df_1$coordinates[i] <- list(c(as.numeric(dum_coord[1]),as.numeric(dum_coord[2])))
}

# print head of scatter_data_df_1
head(scatter_data_df_1)

##               name code            address entries exits
## 1 city liste 1 : 1  CL1     Berlin Germany    5008  3112
## 2 city liste 1 : 2  CL1  Karlsruhe Germany    2002  2223
## 3 city liste 1 : 3  CL1  Stuttgart Germany    3453  3498
## 4 city liste 1 : 4  CL1   Mannheim Germany    2478  3041
## 5 city liste 1 : 5  CL1 Heidelberg Germany    3811  1003
## 6 city liste 1 : 6  CL1  Frankfurt Germany    1875  3135
##           coordinates
## 1  13.38886, 52.51704
## 2   8.40342, 49.00687
## 3 9.180013, 48.778449
## 4 8.467236, 49.489591
## 5 8.694724, 49.409358
## 6 8.682092, 50.110644

I can now create the scatterplot, using the deckgl function from the deckgl R-package.

# define properties of the plot
properties_1 <- list(
  getPosition = get_property("coordinates"),
  getRadius = JS("data => Math.sqrt(data.exits)"),
  radiusScale = 1000,
  getColor = c(255, 153, 77)
)

# plot scatterplot
deckgl(zoom = 10.5, pitch = 35, longitude = 8.40342, latitude = 40.00687) %>%
  add_scatterplot_layer(data = scatter_data_df_1, properties = properties_1) %>%
  add_mapbox_basemap(style = "mapbox://styles/linnartsf/cjq6p9q8f8zwf2rp74qf2o3d5")

We end up with the following scatterplot:

Please feel free to to check out my other posts on spatial data analysis and spatial data visualisation in R.

Linnart Felkl

Data scientist focusing on simulation, optimization and modeling in R, SQL, VBA and Python