Let’s look at the 311 calls on block group level. How is the distribution and are there any areas showing specific spikes when it comes to COVID related calls?
As the first try, let’s plot the count measure of 311 data to simply see how the calls are distributed across block groups.
## Reading layer `count_311_bg' from data source
## `/Users/Shared/Files From d.localized/School/summer 2021/count_311_bg.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 646 features and 15 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -71.19125 ymin: 42.22793 xmax: -70.64807 ymax: 42.45012
## Geodetic CRS: NAD83As suspected, out of 646 BGs, we have BGs with high numbers of calls while there are some other ones with no calls. How would this look like on a map?
The BG “250250303003” lights up red marking the congress street right next to Norman Leventhal Park. That BG only have had the most calls. The sorrounding BGs also show somehow high number of calls. Seaport area and a BG in north dorchester are also among areas with considerable calls.
In making this map, I did not drop the BGs with low population - they also had a lot of calls. There are also several BGs with 0 population but hundreds of calls. Those were left intact as well.
But what if we normalize by population?
## Reading layer `count_311_bg' from data source
## `/Users/Shared/Files From d.localized/School/summer 2021/count_311_bg.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 646 features and 15 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -71.19125 ymin: 42.22793 xmax: -70.64807 ymax: 42.45012
## Geodetic CRS: NAD83 This little addition to the method drastically changed the distribution of calls across BGs. Now we have small number of BGs which are holding a huge number of calls. Let’s map this.
The Boston Common, along with the areas along Chales river (such as Esplanade) are among the light-up BGs. The BG that includes the Marine park also is quite red. In the next map, I will exclude the BGs with population less than 200 - including the places with zero official populations.
This map also shows Franklin park in red - there was population of 380 officially recorded for that BG. As expected, with dropping some of the BGs with top counts, the rest of the city is also showing more colors. We can still see most activities in downtown area, seaport and south.
We can do the same analysis on COVID and Violation measures of 311 data. Followed is the distribution of calls about COVID minus the violation reports.
## Reading layer `covid_words_311_bg' from data source
## `/Users/Shared/Files From d.localized/School/summer 2021/covid_words_311_bg.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 646 features and 15 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -71.19125 ymin: 42.22793 xmax: -70.64807 ymax: 42.45012
## Geodetic CRS: NAD83
We can still see a few BGs with hundreds of COVID calls. Let’s map this.
Again, we can see a couple red ones at Downtown abd seaport. Let’s consider the population of BGs again and see how this changes the distribution.
## Reading layer `covid_words_311_bg' from data source
## `/Users/Shared/Files From d.localized/School/summer 2021/covid_words_311_bg.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 646 features and 15 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -71.19125 ymin: 42.22793 xmax: -70.64807 ymax: 42.45012
## Geodetic CRS: NAD83
We see the same thing happening in COVID measure. When controlling for population, a majority of BGs show zero calls with a high number of calls coming from a few BGs. The following map shows the same measure normalized by population. BGs with population less than 200 are also included.
The Marine park, downtown and Chales river side areas show more reddness. But what if we drop the BGs with population less than 200 once again?
The BG “250250303003” lights up red marking the congress street right next to Norman Leventhal Park just like it did when I mapped the counts of calls. The surrounding area along with the Fraklin Park BG also show a higher rate of calls.
Let’s get to the Violation measure constructed using the 311 data.
## Reading layer `violation_words_311_bg' from data source
## `/Users/Shared/Files From d.localized/School/summer 2021/violation_words_311_bg.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 646 features and 15 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -71.19125 ymin: 42.22793 xmax: -70.64807 ymax: 42.45012
## Geodetic CRS: NAD83
We see the familiar pattern of descending exponential function repeating in the distribution of Violation measure across BGs. Let’s map this one as well.
This one shows some differences. Sounth Boston, Seaport, Downtown and some areas of South End are light-up considerably. Let’s bring in population back into account.
## Reading layer `violation_words_311_bg' from data source
## `/Users/Shared/Files From d.localized/School/summer 2021/violation_words_311_bg.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 646 features and 15 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -71.19125 ymin: 42.22793 xmax: -70.64807 ymax: 42.45012
## Geodetic CRS: NAD83And now let’s map this same distribution:
Seems like not removing the less populated areas skews the maps. Followed is the next try:
The map of Violation measure constructed out of COVID violation reports shows the highest rate of calls in South End, Downtown, South Boston and Franklin Park - as it did in the not normalized map we presented earlier.