Playing with Fire…incidents data. An open data experiment

Following the Blue light camp earlier this month, I’ve been playing with some interesting data on incidents attended by West Midlands Fire Service.

The raw data comprises a list of almost 82,000 incidents attended between 2011/12 and 2013/14, with fascinating facts about each incident: including whether it was false alarm, fire, or road traffic accident; how many appliances attended; and, where appropriate, the type of property involved. There is also important information about the location, including the associated census 2001 neighbourhood (Lower Super Output Area), administrative Ward, and local authority.

That got me wondering whether (a) it was possible and useful to show the data on a map; and (b) I could link it to related third-party sources for equivalent geographic areas.

After some experimentation, I’ve developed and published a basic application at


There are two main elements to this app.

The first is a map with bubbles, sized according to the total number of incidents attended within each neighbourhood (i.e. Lower Super Output Area).  The bubbles are positioned according to the population weighted centroid for each LSOA, which I sourced from the Office for National Statistics’ geoportal.

 The second element is a table which again focuses on the LSOA geography,  bringing together the 2001 Indices of Deprivation ranking with incidents attended, grouped by type of incident: i.e. False Alarms, Fires, Road traffic accidents, and special service call-outs (e.g. water rescues).    Above the table is a scatter chart, tabulating the number of and percentage of false alarms attended – each dot represents an individual LSOA.image01

Clicking on the mapped bubbles will also provide additional data about incidents attended in, and the overall deprivation ranking of individual LSOAs.  So, in the screen shot below we can see the annual trend for incidents of different types in the Birmingham 033F LSOA, alongside the the overall deprivation ranking for 2010.


 What can we learn from this?

 From my perspective, the app is no more than an experiment to:

  • Test and demonstrate potential for blending together related open data sources over the web; and
  • Visualise the results in new/interesting ways using free, open source tools.

 I’ll leave it to you to explore whether there is a correlation between deprivation, and false alarm call-outs.

 Regarding blending data sources,  the app draws directly, in real-time, on the Indeces of Deprivation rankings from DCLG’s OpenDataCommunities service.    I’ve developed some queries which extract overall deprivation ranks for LSOAs in the West Midland Fire Authority area.   If you’d like to know more about how this works, please ask me.

 The app then combines deprivation ranks with data on numbers and locations of incidents.   I’m holding the latter in DCLG’s Geoserver, and querying/retrieving directly over the web via the Geoserver’s Web Feature Server interface.   Again,  please ask me if you’d like to know more about how this works.

 To visualise results, I’m using various free tools and resources, including Ordnance Survey’s OpenSpace API (for backdrop mapping), OpenLayers (to render data on the maps), Highcharts (for the line charts), and JQuery to package and present results in the browser.

I could have gone further, for example, to draw in and visualise additional data sources – such as incidents of anti-social behaviour attended by the Police, and available via the API.  

There is also a similar interesting list of incidents attended by the Greater Manchester Fire Service.   Which gives me another idea….

Posted in Benefits of open data, Informing Decision-making, Posts from feeds, Public safety, Smart communities, Visualising data Tagged with: , , , ,