Data Analysis on The Outliers of Mass Shootings
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Interact with the data here: https://apaciorek.shinyapps.io/shiny2/
Introduction
The debate around gun control and gun violence is emotional and contentious. The goal of this project was to analyze data on mass shootings collected by the Gun Violence Archive to understand the frequency and victim impact of these events. An effort was made to frame this data with as little personal bias and judgement as possible. That being said, every shooting is a tragedy. There are some events that dominate the news cycle, national discussion, and our collective consciousness. My aim in analyzing gun violence data was to put these events into context.
There is no consistent definition of ‘mass shooting.’ Borrowing from the FBI’s definition of mass murder, the Gun Violence Archive defines a mass shooting as any event where four or more people are injured and/or killed.
Data
I analyzed the GVA’s mass shooting data collected from 2014 to 2019. It became immediately apparent that more often than not, these events tended to claim few lives, and events with large numbers of injuries and fatalities were infrequent. Most often, there were zero lives claimed and four injuries per incident, and this was consistent for every year analyzed. Further analysis is needed to understand to what extent the spike at four injuries is a function of the GVA’s definition of mass shooting.


Data Analysis
It is clear from the data that most mass shootings do not claim a large number of lives. Perhaps because of this, they go unreported in the media. It is also clear that the most extreme data points - the deadliest and the rarest - are the ones that are most well known to us. The mass shooting data frames available in the GVA archive do not record the type of firearm or incident, so in order to understand these events, I searched for other collections of information. Joining different data sets was complicated by the fact that definitions are not standardized.
Nevertheless, I hypothesized that perhaps the most common number of injuries per event was four because of the nature of the weapons being used to perpetrate these crimes and the limit on ammunition. More research corroborated this hypothesis: handguns are the most commonly used weapon in mass shootings.

Conclusion
This information is collected and visualized in Shiny. I was interested in studying this data in order to better understand the issue free from the emotional exploitations of all sides of the debate. What I found is that extreme outlier events guide the national conversation, and the data show the problem lies elsewhere. Perhaps a more productive discussion can be had if it were more focused on the most common types of events, and not those on the fringe.
Sources:
Gun Violence Archive: https://www.gunviolencearchive.org/
Statista: https://www.statista.com/statistics/476409/mass-shootings-in-the-us-by-weapon-types-used/
Everytown: https://everytownresearch.org/massshootingsreports/mass-shootings-in-america-2009-2019/