February 13, 2013

The Urban Geographies of Tweets in Africa

This is a quick post containing a few visualisations of information densities in a selection of African cities.

Below, you can find maps of all geocoded tweets published in November 2012 in Accra, Cairo, Dar es Salaam, Cape Town, Johannesburg, Lagos, Tunis, Nairobi, Kigali, Mogadishu, and Addis Ababa.

Look for the information presences and absences; groups of people who are and aren't participating in each city. But also look at the significant differences between cities. Cities like Nairobi, Cairo, and Cape Town are swimming in thick clouds of information, whereas in Mogadishu and Addis Ababa we barely find any digital geospatial information at all. 

If you're interested in why these geographies of information might matter, then check out any the articles at the end of this post. Otherwise, enjoy the maps - and please share any insights you have about any of these cities (or let us know if there are other cities that you'd like to see mapped).

Relevant articles:

Graham, M. 2013. Virtual Geographies and Urban Environments: Big data and the ephemeral, augmented city. In Global City Challenges: debating a concept, improving the practice. eds. M. Acuto and W. Steele. London: Palgrave. (in press).

Graham, M and M. Zook. 2013. Augmented Realities and Uneven Geographies: Exploring the Geo-linguistic Contours of the Web. Environment and Planning A 45(1) 77-99.

Graham, M., M. Zook., and A. Boulton. 2012. Augmented Reality in the Urban Environment: contested content and the duplicity of code. Transactions of the Institute of British Geographers. DOI: 10.1111/j.1475-5661.2012.00539.x  


  1. This is fascinating. Thank you for posting this! Could you map tweets in and around Kumasi, Ghana (Ghana's second largest city), Kampala, Uganda (capital) and Harare, Zimbabwe? Or, is there a tutorial out there on how to do this?

  2. Great visualizations! I'd love to see Douala, Cameroon, mapped, if you ever do a second batch!

  3. I agree, this is impressive! I'd be interested in a map of Bulawayo, Zimbabwe. Is there any way to retrieve this information?

  4. Thanks for sharing! Did you analyse Dar es salaam as well? I think that image might have gotten left off.

  5. Would it be possible to share the methodology whereby this twitter data was geocoded?

  6. Interesting to see the Nairobi map with lots of tweets in most of the populated parts except the Kibera slums.

  7. Thanks for the awesome post, and the papers discussing its relevance. I am curious, though, what technology do you use to query and mine the tweets? Thanks

  8. Nice post! I would like to direct your readers to http://worldmap.harvard.edu/tweetmap, where they can make such maps interactively as well as render heatmaps of the percentage of tweets that contain a query term. TweetMap is powered by a GPU database that I've been building, which can use any number of GPUs to allow queries like these to be run in real-time (we use 4 Nvidia 680s on this system). We eventually should have a billion tweets online for people to search interactively.

    1. Thanks for the link. Why, though, do you limit who can download a CSV of the data (in my case, tweets) that are being explored? Thank you!

    2. Matt,

      Thanks for the reply. Downloading tweets eats up a lot of network bandwidth, so the idea was to limit load on the server by restricting the csv download functionality to those with Havard IDs. If you need geocoded tweets though, please see the page I set up here... http://geops.csail.mit.edu/geo_tweets/ . You'll need to follow the instructions on the page and use the provided script which will set up a partitioned (by day) postgres database and download and restore the dumps for you automatically. Should be about a billion tweets in these files, but you can download any subset of dates you want. I'll eventually put up csv files of the tweets when I get the time. Hope this helps!

  9. Thanks for Posting ! first time I have found a genuine post related to Best ISP in Ghana and west Africa

  10. Hey, Sprawl would be populating the desert with suburbs and cities, however, the overwhelming majority of Clark County is desert that has NOT been populated, just look at a map. Metropolitan areas are delineated at the county level, and the Southwest has huge counties because that's just how someone decided to draw the lines on the map.Thank you!


Note: only a member of this blog may post a comment.