Who is doing what, where visualisation: From concept to use
How open source software, open data and UN OCHA fostering a welcoming community made the 3W dashboard possible
In November 2013 Typhoon Haiyan swept across the Philippines killing over 6000 people and leaving nearly 2 million people homeless. Due to the scale of the disaster the British Red Cross mapping team, was asked to support IFRC response having previously only worked on domestic projects.
In the wake of humanitarian disasters the United Nations Organisation for the Coordination of Humanitarian Affairs (UN OCHA) produces a data set called the 3W detailing who is doing what and where. For Haiyan, UN OCHA decided to release the 3W as open data hosted on Google Spreadsheets. I dived in, but was immediately overwhelmed by the size of the data. Haiyan was a large response with 100s of actors and 1000s of rows of the different activities occurring.
Map Action were producing their very good paper/PDF maps of the 3W at the time. However, static maps meant the data was only sorted by one of the three possible dimensions, the where dimension. In the image below you can see it is easy to read what is happening in a particular area, but harder to see where an organisation (Who) is working or gaps in a particular sector (What).
Our team found that in our domestic work we often got passed data sets that were not always best communicated on just a map. My colleague Andrew Braye had pointed me in the direction of the JavaScript library d3.js and the accompanying dc.js as a possible solution for visualising this type of data.
This is where I saw a use case for an interactive visualisation. A dashboard that would allow the user to filter and display data in a way most useful to them. The principle being:
- click an area on the map and see who is working there and what they are doing
- click an organisation and see where they are working and in what sector
- click a sector to see who is working where for this activity and where the gaps are
This allowed data to sorted, filtered and visualised on any one or more of these three dimensions. The dashboard needed to be able to answer any question about who is doing what and where in a few clicks to instantly render a unique map and charts presenting the answer.
After a couple of days of learning and pulling my hair out I finally got my head around how the libraries worked and produced the first iteration.
I sent this to UN OCHA and it got some traction as a useful tool. It was hosted on humanitarianresponse.info and in total got a few hundred unique visitors.
However, there were still limitations to its use; the biggest restriction being the lack of panning and zooming of the map, but also the bulky file size and the graphs not showing the full depth of the data.
Since then I have produced a few versions of the dashboard (Iraq, Ebola, Cyclone Pam, Typhoon Maysak) for a number of responses with gradual improvements in each iteration solving some of the prior problems. I also released the project as open source on github. Since then, great coders have improved my code or rewrote it to produce some great pieces of work.
The 3W dashboards were gaining a lot of traffic with 1000s of unique visitors, however their creation was restricted to me making them on a per disaster basis or a coder customising one to serve their use case. Andrej Verity contacted me about the possibility of making this tool more accessible. This year I got the privilege to work with the Humanitarian Data Exchange Team and developed a reusable version of the dashboard that could be set up with less effort.
The first version of this was completed a month ago and is already in use for the Nepal earthquake response and also in Somalia. There are still many improvements I wish to make with the end result being a more comprehensive overview of a particular response tying activities to needs in each area. Still, at this point the 3W dashboard is available for anyone to use and visualise their data via the Humanitarian Data Exchange.
This project would have in no way been realised without the open source libraries of d3.js, crossfilter.js and dc.js, the open data released by OCHA, OCHA being open to a community led project and the Humanitarian Data Exchange providing a platform for the humanitarian response community to share and visualise their data.
All versions of the dashboard and HDX component are available on my github account.
Github
