Today was DS18’s first day of Dashboard week.
We were tasked with the challenge of using data from EPA (the Environmental Protection Agency) to build a visulisation that provides a piece of analysis. In my mind this meant creating a more explanatory rather than exploratory dashboard.
So instead of the dashboard having lots of fancy interactivity and cool drill downs, I wanted mine to present a message and back it up with a couple of graphs. Was this the right route to go down? I don’t know but it certainly lead me down some rabbit holes.
The data as it was provided contained multiple CSV files all of which gave information on different facilities the EPA monitors. These facilities include anything from factories to universities. By joining some of the CSVs provided together you could build a table of all the facilities across the USA and what kinds of environmental interest EPA has in those facilities.
Whilst looking at this data, I noticed that one of the fields provided gave information on whether the facility was in tribal land or not. Having heard lots on the news about struggles between native Americans and oil companies in Alaska, I thought it could be interesting to investigate whether there was a difference in toxic material release in and out of tribal land in Alaska.
This led me down a wee rabbit hole. I managed to fine some brilliant toxic material release data for which our original data set was purpose built to be able to join to. I then started doing some mapping and noticed that many of the points which were not labeled as being in tribal land clearly were in tribal land (according to a US census shape file for tribal land). I know that other aspects of the CSV file’s spatial data were a bit dodgy (e.g. state names and state codes not matching one another), so I decided to build my own boolian ‘Tribal land’ field using the spatial matching tool in Alteryx with the shape file provided by the US Census Bureau.
Once I’d done all of this and thrown it into tableau I realised that so little of the facilities were in tribal land, the analysis would’t be great.
So instead I had to find another story in the data.
Eventually I discovered the trend shown on my dashboard – that since 2010 the amount of lead released in California has been going up every year.
In the end I had to build a rushed dashboard that I’m not very proud of at all really… But hey ho there’s always tomorrow.
Link for viz bellow: