Research Question: Have Fireworks Complaints Gone Up During the COVID-19 Pandemic?
Based on the anecdotal experience of myself, family, and friends, it seems like personal firework usage (all of which is illegal) has gone up since the COVID-19 pandemic began. So I have filtered NYC OpenData’s 311 Service Requests from 2010 to Present on complaint type (include: “fireworks”) to see if their data supports this observation. Assuming fireworks are most prominent in the summer around the 4th of July holiday, I have limited the search data to start from March 1, 2018 (two years before the pandemic began in NYC) and end on August 31, 2021. This includes two summers per-pandemic and two summers during the pandemic.
Audience
Fireworks of all kinds are a major concern for public health–both mental and physical. The explosions can be triggering for people experiencing post-traumatic stress disorder (see: https://www.pennmedicine.org/news/news-blog/2020/july/the-overlooked-effects-of-fireworks). They can also cause undo stress in animals; according to the American Kennel Club, more pets go missing July 4th and 5th compared with any other time of year (see: https://www.akc.org/expert-advice/health/prepared-pets-go-missing-july-4-5-day/). Furthermore, a first-of-its-kind study published last year (see: https://nyulangone.org/news/common-fireworks-release-lead-copper-other-toxic-metals-air) shows that the airborne chemicals released from fireworks are much more toxic compared with everyday air pollutants we breath in. My audience is public health officials, as well as New Yorkers concerned with the health and wellbeing of themselves, their families, and fellow New Yorkers.
Description of the Visualizations
I first created several density maps to visualize fireworks complaints across the five boroughs. In the dashboard below, the dot density map on the top shows every complaint within this given time period. The tooltip shows the latitude and longitude of the complaint, the date and time the complaint was made, and response to the complaint (it seems these complaints were exclusively addressed by the NYPD), and the status of the complaint (all of them seem to be “closed”). The choropleth map shows the density of complaints for this time period by borough. Brooklyn had the most complaints, followed by Manhattan, Queens, the Bronx, and lastly Staten Island. The tooltip shows the total number of complaints in each borough. In creating a line chart showing the trends of complaints (see the second dashboard), I realized that June and July are the months with the most complaints. As such, I created another dot density map that only shows the complaints for each of the year’s month of July. The tooltip provides the same information as the other dot density map. You can see that July 2020 has the most coloring on this map, but to help make this even more clear, I created a companion bar graph that shows the totals for each of the Julys, with the same color coding as the corresponding dot density map.
In the second dashboard, I charted the overall trends in the numbers of complaints made to 311. Indeed the spike in calls in June and July of 2020 (and to a lesser but still large extent June and July of 2021), make the spikes in calls for the same months in 2018 and 2019 almost impossible to see in this line chart. The tooltip shows the total number of complaints each month. The bottom line chart shows the same data broken down by borough. The trend is consistent across all boroughs for this time period.
In the third dashboard, I have explored this data by different types of locations. There is a “Location Type” field in the 311 data, so I first created a bar graph to show which location types had the most complaints. Not surprisingly, based on how much firework debris you find in the streets after the 4th of July, “Street/Sidewalk” is by far the most common location type where illegal fireworks are complained about. The second bar graph shows the total number of complaints per borough, color-coded to match the borough line chart in the second dashboard. Complaints also are generally assigned a zip code, so lastly I graphed the total number of complaints by zip code, the bars of which were further color-coded by borough.
Data and Design Decisions
There is only one complaint type: “Illegal Fireworks.” According to 311, “In New York City, all consumer fireworks, including sparklers, are illegal to use, buy, sell, or transport” (see: https://portal.311.nyc.gov/article/?kanumber=KA-02249). Almost all of the visualizations above rely on on the “Unique Key” for each complaint, but it is possible that multiple people could have called 311 about the same incident. It’s also possible that some incidents went without complaint as in a large population, everyone may have assumed someone else had already called it in, or some may have issue with reporting such things in general. There were a few null values that I excluded from the visualizations. The most prominent of which was in the zip code graph for which there were many complaints assigned a zip code but not a borough.
I originally only planned to create one dot density map, one line chart, and one bar graph, but as I started seeing patterns with some of the visualizations, I decided to play around more to see what else I could show, and how best to show that data. I like the dueling dot density maps, but I was worried the layering of colors in the Julys one made it somewhat harder to discern the differences, so I created the companion bar graph to more easily display this information. Similarly, when I saw such a clear trend in the months of June and July, I wanted to see if this trend was consistent across boroughs. For each visualization that is meant to show data by borough, I chose a consistent color theme from the “color blind” palette in Tableau. The only exception is the choropleth, for which I chose an orange gradient to be consistent with the dot density map next two as these two maps are both showing density. For all other visualizations, I chose different color coding (avoiding mixing red and green) that I thought provided a high contrast between data points and/or made it clear that this visualization was different from the borough visualizations. Grouping together visualizations with the same color schemes also meant that I could include fewer legends and give myself more space for the visualizations and accompanying captions in the dashboards.
Next Steps
Within this current dataset, I would be like to go in and clean up the data. For example, there are fields that are empty that seem like they could easily be filled (i.e., complaints with a zip code but no borough associated with it). If there are several zip codes without a borough, how is this affecting the rest of the borough visualizations?
I would also like to run some kind of text analysis on the “Resolution Description” field to see what trends are there in the responses. How often are tickets closed when there isn’t really a resolution (e.g., “The Police Department responded to the complaint and with the information available observed no evidence of the violation at that time.” seems very common).
Finally, I would be interested to compare this dataset with previous years. Were 2018 and 2019 exceptionally low? Are there other years where fireworks complaints have skyrocketed (the 400th anniversary of Henry Hudson immediately comes to mind, though this was in 2009 and predates the available 311 data).
