My approach with this week's assignment to construct some initial quantitative views was to focus on current census data, obtained through the US Census's API. I used the American Community Survey 5-Year Data (2017)'s Data Profiles endpoint, which "contain[s] broad social, economic, housing, and demographic information. The data are presented as population counts and percentages. There are over 1,000 variables in this dataset."
I focused this week's assignment on five particular statistics, all compiled at the census-tract level:
- % of the population who own their home
- % of the population whose income in the last 12 months was below the poverty level
- % of the population who are nonwhite
- Median market value of homes
- % of homeowners who have a mortgage
I thought these statistics might tell an interesting story on the current legacy of redlining policies on the areas they encompass, but I want to make sure I stress test this list (and also add wealth metrics to the list).
The first views I constructed were map-based, with each of these metrics plotted on separate maps by census tract. I then overlaid HOLC rating visual filters on these maps, allowing us to see a subsection of this overall map based on redlining grades. Interactive prototypes for those maps are available at http://ryanabest.com/ms2-2019/quant/ (these prototypes maps take a little time to load and may need to be refreshed once when accessed).
I also wanted to look at this data outside the context of maps for easier comparisons from zone to zone and tract to tract. To do this I used the shapely python library to crudely estimate which HOLC zones each census tracts overlaps with. Many census tracts overlapped with multiple HOLC zones with different grades, so for now each match of tract-to-grade is retained (i.e. if a tract matched with a 'A' and a 'B' zone, it will get two records in my data set). I then constructed some rough-draft views in Tableau to compare these metrics across tracts based on their HOLC grade. Overall, it looks like A-Rated zones still have higher proportions of white homeowning residents than their lower-rated counterparts.