In dgl5gw/geocompr: Geocomputation with R: a book

knitr::opts_chunk$set(collapse = TRUE, comment = "#>")

Reading vector data

library(sf)
vector_filepath = system.file("shapes/world.gpkg", package = "spData")
world = st_read(vector_filepath)

To demonstrate this, we will use a function to compare st_read() with its sp equivalent, rgdal::readOGR():

bench_read = function(file, n) {
  m = microbenchmark(times = n,
                     rgdal::readOGR(vector_filepath),
                     st_read(vector_filepath)
  )
  mean(m$time[1:n]) / mean(m$time[(n + 1):(n * 2)])
}

This function takes as arguments an input file (file) and a number of times to run each command (n) and returns how many times faster st_read() is than readOGR(). Let's run the benchmark for the world.gpkg file represented by the object vector_filepath:

library(microbenchmark)
read_world_gpkg = bench_read(file = vector_filepath, n = 5)

read_world_gpkg

The results demonstrate that sf was around r round(read_world_gpkg) times faster than rgdal at reading-in the world countries vector. The relative performance of st_read() compared with other functions will vary depending on file format and the nature of the data. To illustrate this point, we performed the same operation on a geojson file and found a greater speed saving:

vector_filepath_gj = system.file("shapes/cycle_hire_osm.geojson", package = "spData")
read_lnd_geojson = bench_read(file = vector_filepath_gj, n = 5)

read_lnd_geojson

In this case sf was around r round(read_lnd_geojson) times faster than rgdal.

Writing vector data

library(rgdal)
world_sp = as(world, "Spatial")

system.time(writeOGR(world_sp, dsn = "world_sp.geojson", layer = "world_sp.geojson", driver = "GeoJSON"))
system.time(writeOGR(world_sp, dsn = ".", layer = "world_sp", driver = "ESRI Shapefile")) 
system.time(writeOGR(world_sp, dsn = "world_sp.gpkg", layer = "world_sp.gpkg", driver = "GPKG"))
system.time(st_write(world, "world_sf.geojson", quiet = TRUE))
system.time(st_write(world, "world_sf.shp", quiet = TRUE)) 
system.time(st_write(world, "world_sf.gpkg", quiet = TRUE))

world_files = list.files(pattern = "world*")
file.remove(world_files)

dgl5gw/geocompr documentation built on May 18, 2019, 8:11 p.m.