As mentioned in my previous Lab, I felt as though the Miami-Dade 311 Dataset would be ideally displayed as mapped information. I titled this map 311 MIami-Dade 311 by Zips because it shows the amount and type of 311 calls in zip codes of Miami. During this lab, using Carto, I used new columns from that dataset in which I mapped using shape files from the cities open data site. I was able to create an interactive map that helps understand the 311 data better. The results of this map also expose more evidence to what I found using Tableau, in which I concluded that there are inconsistencies with how the 311 data is collected.The questions I aimed to answer through the map were:
- Which zips codes utilize the 311 help line the most?
- What neighborhoods do the zip codes fall into?
I feel like when it comes to interest in data visualizations, I’m most interested in those that are maps. I like maps because I feel as though they have the ability to display a good about of different information, and are easy to compare and contrast with each other, and when done right they’re easy to read. The first map vis I thought of as inspiration is one created in 2017 and published by the New York Times. These choropleth maps utilize Youtube’s geocode streaming data to show the relative popularity of popular musicians in different parts of the country. The maps use a light to dark scale to show variations within what I believe are counties (fig.1).
Another inspiration of mine is a map that was featured in an exhibition at Pratt Manhattan Gallery titled The Asthma Quilt (fig. 2). The quilt, created by Perkins+Will/Quilters without Borders is a three dimensional quilt that is a choropleth map of some of Manhattan and most of Brooklyn, then broken down to what I also believe are counties. The map shows which areas of New York are most effected by asthma. This map too uses a gradient of light to dark. Light meaning there are less cases of asthma, and the darker/red to mean more. The artists use grey embroidery thread to mark the location of co-ops in New York, and red plus signs to identify hospitals, thus the map is addressing poorer health conditions in lower income areas. One thing I also like about this map, is the use of green and blue fabric swatches to help identify particular landmarks such as parks and bodies of water.
fig.3 Perkins+Will / Quilters without Borders, The Asthma Quilt, 2016–17. Recycled wool felt fabric samples, embroidery, 60 x 80 inches.
My third inspiration, a map that was shared with me recently, is potentially a direction I can see my map aesthetically going in the future. Jill Hubley’s map Victims of Police Shootings in the US is a map true to its title. This timely choropleth map maps police shooting deaths from 2013 – March of this year 2018. This map outlines each state then uses a gradient from light to dark to represent the density of white people in each county of the US. The map then like mine, uses dots to show all the individual cases. When you hover over a county, the space lights up a brighter color, and information from the US Census data about the demographics of the counties population is displayed. She uses different colors dots to represent the race of the victims and when you hover over the individual dots, you can see the victims name (if known), date of death, race, gender, and county of death. This map is very informative once you understand the gradient scale, and really illustrates some interesting findings regarding police shooting deaths.
fig.4 Jill Hubley Victims of police shootings in the US
For this map I used the Miami-Dade 311 dataset found on the Miami-Dade County’s open data site. I was also able to find a website dedicated to spatial data, the “Miami-Dade County’s GIS Hub,” which was very helpful when searching for ways to illustrate my map. The GIS Hub states that it “promotes mapping technology as a tool used to better understand our community.” The site is extremely easy to search and explore different categories, such as building, education, electoral, and hydrology. The site is user friendly by providing map suggestions, and interactivity directly on the page, so that users can test the map before downloading. The site also provides two APIs for users to copy into their mapping software.
For mapping the datasets I used Carto, which made it very easy to input the county provided GeoJSON links from the GIS HUB, and also my 311 dataset.
After creating a login for Carto, the software easily downloaded and plotted points based on the inputted 311 dataset. There were some null plots that appeared to be floating in the middle of the ocean on the map. To resolve the floating points we altered the longitudes using Carto instead of trying to locate all the incorrectly inputted locations within the dataset. After this step and some trial and error, all the plots viewable were in Miami-Dade County.
After inputting the Miami-Dade 311 data and fixing the longitudes in Carto, I found the Miami-Dade GIS Hub. On the hub I started by exploring what shape files they had, and identified what polygon feature class would be most informative for my questions. Since I was aiming to answer the questions of which neighborhoods use 311 the most, I knew I was looking for a polygon feature class that showed the counties, districts, and/or neighborhoods of Miami-Dade County. I ended up utilizing a polygon feature class of Miami-Dade County Zip Code boundaries. The attributes in this spatial dataset were zip code id, and the shape and length of the boundaries. I also used a dataset of the Miami-Dade Municipal Boundaries. This data included the municipal codes and neighborhood names that people are most likely to know and recognize when referring to Miami-Dade County.
Once the three datasets were all inputted on Carto, I stylized the map. The first style choice I made was having the 311 calls displayed as dotted plots. Therefore they looked as if they were individual occurrences of calls. I also wanted the map to be able to show and easy overall interpretation of the 311 calls per area. From the Tableau lab I learned that the 311 data captured zip codes more so than the accurate neighborhood names. Therefore I decided that I wanted the zip code shape file to be a choropleth map that would be a gradient of light to dark. Light referring to lesser 311 calls and dark meaning more calls. Even though from the zip codes you could see where the most calls were coming from, I also wanted people to be able to identify which neighborhoods those zip codes resided in; since as mentioned before generally people refer to the neighborhood and not zip code. Therefore I decided to use a red outline to show the more familiar neighborhood boundaries.
For all the data layers I used click popups so that users could get information such as the neighborhood, zip, and issue recorded for that plot. I decided to use the Positron basemap on Carto because it’s light in color, and the streets and other finer details weren’t important for my map. I want people to be able to see the primary story I’m telling, which is which neighborhood or zip codes, in this case report the most issues using 311. I used the a burgundy gradient scale with 5 colors that map out the density or amount of calls per zip. Once people understand that, they can click on the individual dots and see what the issue is that was reported at that point. I added clearer titles, a legend, and widgets to my map so that users easily understand what they’re looking at, and are able to explore the different issues reported and where they’re located.
My Carto map is successful in what it shows, which is which zip codes received the most 311 calls. Though the map also draws attention to the lack of calls in particular areas and neighborhoods. It leads me to wonder if those areas truly aren’t using 311, and if so why? Or is there really something up with the way the 311 data is collected. On my map you can see there are huge gaps of white that are illustrating entire counties that have no 311 data (fig.6). At first I thought that the maps on the Hub were iffy and leaving out particular neighborhoods. But after recreating the map, I figured out those areas that are white and appear to not be apart of the county, have no recorded data.
I’m also curious into understanding what and why particular areas in Miami are considered to be in the municipal boundaries, and others are not? They’re not only unmapped on the shape files downloaded from the city, but the calls also seem to be mapped around the borders of these areas. Even if one is unfamiliar with Miami’s landscape, it’s easy to see that the white spaces are municipal areas.
I find that the Issue Reported widget is helpful, though the word “other” displayed in the widget is unclear, and using the search on the widget is not helpful to someone who is unfamiliar with the data. Therefore I guess you could really say the map is unsuccessful because it’s not 100% true to the 311 data. There are way more “other” calls that appear not to be plotted.
What I find works about my map, is the consistency in the colors. I think that using the blueish purple for individual calls is clear, direct, and stands out against the gradient burgundies. I also think the red outline of the municipal boundaries is helpful, as those boundaries are a lot more familiar to a native than the zips. For the popups on the individual calls I used a different color than the popups for the zips and neighborhoods. I think this is helpful in identifying what in particular users are clicking on. I also gave the calls a multi color popup that matches the colors of the dotted plots. Therefore the two seem related. Along with using the same color purple in the widget with the issues reported (fig.7).
I think that there are a lot of directions this data can go in in the future. I personally would like to understand why certain areas are not featured on the map. Are those particular areas are not using 311, is it under advertised, or not offered in certain parts of Miami-Dade and why? This could possible be achieved by using Open Refine to clean up the data and create grouping much like the Tableau lab.
Aesthetically I’d like to see different color dots for each type of call, like Jill Hubley’s map in which the different color dots represent individuals of a different race. I’d also like to utilize her hover feature in which the county turns a pale yellow, therefore individuals can differentiate what they’re hovering over.
I played around with seeing this data displayed using the time feature. I thought that worked well with map, because it kind of displayed like calls being pinged all over. I’d love to explore that feature more and understand what time stamps the data is actually portraying, to see if that feature really does speak for the dataset.