The first map posted here, and my first original map experiment with Carto, shows the location of housing developments in San Francisco between 2005 and 2015. The point data show how many affordable units were included in each project, and they are placed against a choropleth background map showing the net number of units added in each of the city’s Census tracts over this period. The data is sourced from the SF Planning Department’s Housing Balance Report, which details the location and affordability breakdown of housing developments over time, sorted by neighborhood and supervisory district, along with brief descriptions of the projects. The mapping of this data shows where housing development and affordable units have been concentrated over roughly the past decade, which appear to be fairly strongly correlated. By including “net units,” it sheds light on the fact that over the course of the development process, units are both gained and lost – lost to replacement, consolidation, and outright demolition. This might explain why in some Census tracts the net change is as few as 3 housing units. It also shows the sprawling nature of housing development in much of the city; large swaths of the western portion of San Francisco show a considerable number of projects, but this has resulted in relatively few total new units. Hopefully this helps the viewer better understand the geography of recent housing development in the city, which has clearly been extremely uneven for a variety of reasons.
Process-wise, I began with some data cleaning that was primarily to prepare the data for geocoding. This took some tinkering with the .csv file and Carto, which was a bit finicky in recognizing the geodata in the table. After uploading this data, I grabbed a shapefile for California Census tracts. I then performed a spatial join between the “net units” data from the planning department and the Census map, which allowed me to create the choropleth in the background. I then used another choropleth, this time in point form, for the affordable units category of data. I also included clickable infowindows to give the precise number of affordable units and net units per project, as well as a description of the project and the affordability income targets.
For the second map, I’m once again using a choropleth, this time to show the location of selected building permits by Census tract in Oakland. This map is admittedly a little more ambiguous about what story it tells, but what I tried to do was take residential building permit data for the years 2003-2013 from the Oakland Open Data portal and isolate only those permits that involved demolition of some kind. My method was pretty crude: I merely put a filter on the “description” column in the original data source and kept only those entries that included the words “demolish” or “demolition,” which is by no means an exhaustive or even accurate way to produce the desired slice of data. This left me with about 1,300 data points out of an original set of over 120,000. Performing a spatial join with the Census tracts resulted in the base choropleth map, which shows, as one might expect, that much of the issued permits over the span of time involving demolition of some kind are concentrated near the port and the industrial stretch along the Bay near the airport. I then brought in a shapefile for Oakland’s Priority Development Areas to see if a pattern might emerge. My thinking was, with any attempt to spur urban development (PDAs are targeted by the State for infill development opportunities, typically near transit), there is also a fair amount of destruction and clearing. The resulting map shows that the tracts with the highest concentration of demolitions are within or adjacent to PDAs, but this is by no means conclusive. The choropleth data also doesn’t take density into account, so it’s hard to glean too much from this. I do find this to be a pretty interesting data set, so it could be something I return to in the future and attempt to dig a little deeper into. Reading through the descriptions of demolitions reveals some interesting stories, so it would be neat to figure out a way to better incorporate that qualitative data in the future.