This weekend I was interested in exploring Utah’s “Open Data Utah” data portal hosted via Socrata. The result of the effort is a tool exploring Salt Lake County’s budget. Some of the writing included here, as well as the subject matter (the budget visualization tool) have been cross-posted on the Code for America Tumblr page for the Salt Lake County 2016 Fellowship team.
Open Data Utah has a plethora of solid datasets that can prove fun to spend a morning or evening with. Data is available, also at the State and City level. For example, this Salt Lake County Budget Explorer utilizes data that is in fact a subset of the overall budget dataset that comprises budget expenditures across all county bodies in the state. All this information was made possible through an interesting initiative by the State of Utah, Transparent Utah.
The resulting tool, called “Salt Lake County Budget Explorer” and available here, is purely intended as an initial exploration into the County budget. I would caution against using it as an absolute resource for a variety of reasons, not least because I believe that Utah does not provide an assurances regarding the Socrata data that is being published. Nonetheless, that caveat aside, I am also presenting this tool as a representation of how I interpreted the Socrata data. I do not claim to understand the language and naming conventions of the budget and there is the possibility I have incorrectly represented values. I created a nesting “3-tier” structure for the chart as I saw that there were a small number of funds and it appeared that, within that, there were overall account expenses each of which had a specific name (regarding that individual transaction). Because names were common and repeated, I was able to then cluster names (the third ring). When looking at the third ring, you will see that some components are not completely full. The reason for this is that there are a number of very small purchases that were not rendered. Essentially, below 0.1%, the cost does not register on the budget visualization tool.
Looking at the code for the project: The project was a simple re-use of a standard D3.js visualization method called “Starburst.” The three tiers happened to coincide with the observed pattern of the budget. This coincidence allowed me to utilize much of the code for the D3 example and thus allowed me to complete the project quickly - in the span of a morning. The code for the project itself lies within the Code for America Team Salt Lake County 2016 organization. Thus, in order to render it under my URL, since it uses my Github account’s gh-pages, I just inserted an iframe into a dummy repo on my account. The code for the actual data visualization tool is, thus, available at it’s repository within the Salt Lake County 2016 organization.
Further discussion regarding Transparent Utah: Administered by the Department of Finance, the site provides a nice search feature to explore all expenditures performed by any County. Looking at Salt Lake County, specifically, one can see that there are a total of 757,995 registered transactions, totaling $1,035,778,082.26. This same information has effectively been downloaded and loaded (cross posted, if you will) into Socrata. I’m unclear on the frequency of update to this tool, as 2015 only contains 293,059 results, suggesting that perhaps last year has yet to be completed (if we are to assume that expenditure count is similar year to year and that there was not a substantial drop in spending from 2014 to 2015 - which is currently listed at $827,148,981.46).
Looking ahead: This project was simply a short exercise in exploring the data that had been released on Socrata. Assuming the 3-tier methodology does indeed make sense for users of the visualization tool, a few key additions should be pursued. Primarily, the improvements would lie in cleaning up the right legend. It is not too useful as it stands right now. Outer rings could be accordian folded into parents tiers. Also, clicking on items in the legend should highlight their counterparts on the ring chart.