This post gives an example of creating maps in R. The full code can be found on Github.

The source data is going to be my gmail archive, not something that seems to be immediately relevant for a map-based project. The idea is to take the origin IP address from each mail, convert to a geolocation and put together a plot showing where all the mails I’ve received have originated.

Formal Concept Analysis Background

The starting point is to extract the features that are required from the email data. The archive is downloaded in a .mbox format and to use it in R, a .csv file will instead be generated. A simple python script does this:

import mailbox
import csv

mbox = mailbox.mbox('All mail Including Spam and Trash.mbox')

writer = csv.writer(open("Mail_Archive.csv", "w"))
writer.writerow(["Subject", "From", "Date", "Labels", "Importance", "Received", "Thread-Topic", "Has-Attach"])

for message in mbox:
	try:
		writer.writerow([message['subject'], message['from'], message['date'], message['X-Gmail-Labels'],\
		message['Importance'], message['Received-SPF'], message['Thread-Topic'], message['X-MS-Has-Attach']])
	except (UnicodeDecodeError, UnicodeEncodeError):
		print('Unicode error!')

This is simply loading the mbox object (Thanks to the mailbox library) which is a list of dictionary items. Iterating over this list, the relevant fields can be extracted and written to the csv file.

The next step then was to load into R and extract the IP addresses from this data.

Another detour into python takes the ip addresses and returns the geolocations.

Creating the map

Next step, start creating the map! First, we want to get a count on the occurrence of each location and set up the map background with ggplot.