A way to simplify package use on a R project

Introduction

One of the steps that I always do when I begin an R script is to check if all the needed packages are installed, and if not install them.

That way I know that the script won't crash on another computer because of a missing package (one less reason to crash :D).

Lately I've been thinking about the subject of code reuse and decided to transform my block of code in to an R function.

The Code

#«««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««
#  script : loadlib.R
#  author : Vitor Chagas(VC)                last updated : 2013.08.19
#«««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««««

# install & load necessary packages ----
#<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<

loadlibs <- function(...) {

  necessary <- as.vector(as.character(match.call()[-1]))

  installed <- necessary %in% .packages(all.available = TRUE)

  # install
  if (length(necessary[!installed]) >=1)
    install.packages(necessary[!installed])

  # update
  update.packages()

  # load
  for(pkg in necessary)
    library(pkg, character.only=TRUE)  

  rm(necessary, installed, pkg)
}

Final comments

Now I just save this code into a lib folder, and then use the following code to source every code file in there

sapply(list.files(pattern="*.R", path="lib", full.names=TRUE),
       source,.GlobalEnv)

Mapping Earthquakes Over the Past 30 Days

Introduction

After reading this post I decided to try it and adapt to my own needs.

This is just a simple example that uses real-time data from the U.S. Geological Survey.

This shows the location of earthquakes with any magnitude over the last month around Puerto Rico (just had to pick some place with earthquakes!).

The Code

#===========================================================================
# project : aportugueseactuary.blogspot.pt
#  script : Mapping_Earthquakes_Over_the_Past30_Days.R
#  author : Vitor Chagas
# updated : 2013.08.10
#===========================================================================

#---------------------------------------------------------------------------
# reset workspace
#---------------------------------------------------------------------------

rm(list = ls())
gc()

#---------------------------------------------------------------------------
# create & set work directory root
#---------------------------------------------------------------------------

# change as needed
workdir <- '.'

dir.create(workdir, recursive=TRUE)
setwd(workdir)
rm(workdir)

#---------------------------------------------------------------------------
# create folder tree (needed folders)
#---------------------------------------------------------------------------

#dir.create('data')

#---------------------------------------------------------------------------
# install & load necessary packages
#---------------------------------------------------------------------------

necessary = c('ggmap')

installed = necessary %in% .packages(all.available = TRUE)

# install
if (length(necessary[!installed]) >=1) 
  install.packages(necessary[!installed])

# load
for(pkg in necessary) 
  library(pkg, character.only=TRUE)  

rm(necessary, installed, pkg)

#---------------------------------------------------------------------------
# get earthquake data
#---------------------------------------------------------------------------

file.url <- c(
  url='http://earthquake.usgs.gov/',
  path='earthquakes/feed/v1.0/summary/',
  file='all_month.csv'
  )

eq <- read.table(file=paste(file.url, collapse=''),
                 fill=TRUE, sep=',', header=TRUE)

# rename columns
names(eq)[2] <- 'lat'
names(eq)[3] <- 'lon'

# discretize magnitude into 5 intervals
eq$mag.size <- findInterval(eq$mag, c(0,2,4,6,8,10))

#---------------------------------------------------------------------------
# generate map
#---------------------------------------------------------------------------

# map center (some place with earthquakes)
center <- geocode('Puerto Rico')
location <- c(center$lon,center$lat)

# get map from Google
map <- get_map(location = location, zoom=7, maptype='roadmap')

# display earthquake data
ggmap(map, extent = 'device') +
  geom_point(data=eq,
             mapping=aes(x = lon, y = lat, colour=mag)) + 
  scale_colour_gradient(limits=c(0, 9.9), low="yellow", high="red")

Conclusions

The process is very simple, but sometimes it needs some fine tune of the zoom and maptype options to get the intended result.

This data source is not enough for my own needs, it's not picking the earthquakes around Portugal, so I need to find an alternative.