Tableau Public Lab Report by Kyle Palermo, June 23, 2020.
Background
Criminal summonses are one of four primary enforcement actions at the NYPD’s disposal (in addition to what happens off-the-books or results in death). Rather than protecting life or property, a plurality of these summonses are issued for marijuana possession or various alcohol consumption offenses (excluding the highest single offense category, federal motor vehicle violations). Looking at criminal summons data for 2019, this visualization shows that the NYPD pays a surprising amount of attention to marijuana and alcohol consumption and that their efforts to suppress these rather ordinary vices target specific neighborhoods and, overwhelmingly, people of color. Police murders remind us that no matter how innocuous the pretext, any encounter with law enforcement can turn deadly, making this small war against victimless non-crime all the more troubling.
Process Documentation
Tableau Public is a free, cloud-based application for filtering, analyzing, and visualizing data, and is capable of handling large data sets that might choke more vanilla spreadsheet applications like Excel or Google Sheets. Rather than importing a data source, Tableau links to it and provides an interface for previewing and filtering down rows to quickly mute extraneous data without altering your source file. From there, users can create additional worksheets to explore and visualize the data, including the standard array of charts familiar to spreadsheet users and some more advanced visualizations such as packed circle charts and tree maps. Users can then aggregate their visualizations into dashboards, and all of their work is housed as a single workbook which can be saved to a profile on Tableau Public’s online community. The public version pens users into Tableau’s proprietary environment, and exporting data or visualizations to other formats largely off limits.
Before beginning my visualization I reviewed projects featured on Tableau Public’s Viz of the Day/Gallery including “Black Owned Eateries in the US,” “International Astronauts,” and “The Color of COVID-19,” as well as Tableau’s how-to resources.
Results and Discussion
My Tableau project includes eight worksheets and one dashboard visualizing the NYPD’s distribution of criminal summonses. The source data is saved from the City of New York’s open data portal, which provides a variety of open-source data sourced from various city agencies for free public download. The first worksheet sums each distinct offense documented in the data and sorts them from most to least prevalent. I then manually identified and grouped categories that describe alcohol consumption and marijuana possession, and filtered out federal motor vehicle safety code violations, which represent the single largest category of offenses but skew the visualization away from street-level law enforcement. A subsequent sheet compares the number of summonses in 2019 for alcohol, marijuana, and all other offenses, using a tree map to maximize visibility and economize on space in a dashboard view. Three subsequent maps show the geographic distribution of summonses in each of these categories across the five-borough area, with dots set to a low opacity to provide a heat map effect wherever summonses are highly concentrated. Three further visualizations break summonses down by race of the recipient (using the NYPD’s race categories but combining “White Hispanic” and “Black Hispanic” into a single category). A dashboard compiles each visualization into a three-column display.
Reflection
My visualization succeeded in manipulating the data and forming it into a legible story, however it would benefit from a more holistic design and more visual development. Rather than breaking up my visualization into distinct chunks, I think the story would be better told on a single map with accompanying charts. Also, the visuals are polished but very basic. Tableau Public’s gallery shows that more advanced visualizations are possible but I frequently hit roadblocks while attempting to manipulate my dashboard and often imagined myself needing to redraw the visualizations in illustrator. Rather than try to hack together a more visually designed presentation I decided it would be best to learn more about properly manipulating visuals in Tableau and then return to it later.
This visualization would also be improved through a more comprehensive approach to the data, including coding the more than 550 distinct offenses to remove all moving violations–rather than only those that are covered under the federal motor vehicle safety code–and to more systematically include/exclude the less common alcohol-related offenses in the data’s tail. Also, criminal summonses are just one part of the picture. Arrests and desk appearance tickets, civil summonses, and stop-and-frisk data are all available to help tell a more complete story about predatory police behavior, but would require a larger effort to harmonize and combine several different data sets. I discovered that neither Tableau nor OpenRefine offer any simple, automated process for joining distinct data sets, so I tabled any thoughts of a more comprehensive data set.
As an Excel native, I often find Tableau difficult to navigate and catch myself searching through forums to figure out how to complete even simple tasks like changing the contents of data labels. But, despite my frequent head scratching, I’ve generally acquiesced to Tableau’s way of doing things and find it to be a very effective tool for taking control of large data sets and creating visualizations in a few clicks.